Front Cover
 Our Objectives
 Appendix A: The Ephemeral Cities...
 Appendix B: Authority Records for...
 Appendix C: An Alternative Interface:...

Title: Ephemeral Cities : final project report
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00082559/00001
 Material Information
Title: Ephemeral Cities : final project report
Physical Description: Book
Language: English
Creator: Kesse, Erich J.,, 1959-
Affiliation: University of Florida -- University Libraries -- Digital Library Center
Publisher: Erich Kesse
Publication Date: 2008
Spatial Coverage: United States of America -- Florida -- Alachua County -- Gainesville -- University of Florida
Coordinates: 29.6508338410016 x -82.3421102043672
 Record Information
Bibliographic ID: UF00082559
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: Available under Creative Commons license: Attribution + Noncommercial + ShareAlike (by-nc-sa).


This item has the following downloads:

PDF ( 7 MBs ) ( PDF )

Table of Contents
    Front Cover
        Page 1
        Page 2
        Page 3
        Page 4
        Page 5
        Page 6
        Page 7
        Page 8
        Page 9
        Page 10
        Page 11
        Page 12
        Page 13
    Our Objectives
        Page 14
        Page 15
        Page 16
        Page 17
        Page 18
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
        Page 29
        Page 30
        Page 31
        Page 32
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
        Page 46
        Page 47
        Page 48
        Page 49
        Page 50
        Page 51
        Page 52
        Page 53
        Page 54
        Page 55
        Page 56
        Page 57
        Page 58
        Page 59
        Page 60
    Appendix A: The Ephemeral Cities Road Show
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
        Page 66
        Page 67
        Page 68
        Page 69
        Page 70
        Page 71
        Page 72
        Page 73
        Page 74
        Page 75
    Appendix B: Authority Records for information extraction and query
        Page 76
        Page 77
        Page 78
        Page 79
        Page 80
        Page 81
        Page 82
        Page 83
        Page 84
        Page 85
    Appendix C: An Alternative Interface: A Google Earth Experiment - Key West
        Page 86
        Page 87
        Page 88
        Page 89
        Page 90
        Page 91
        Page 92
        Page 93
Full Text


Ephemeral Cities
final project report
to the
Institute for Museum and Library Services

submitted by:
Erich Kesse
Digital Library Center
University of Florida Libraries


final project report page 2

Introductory Remarks
and a Foreword

In 2003, the University of Florida's Digital Library Center set out to do
something never before done: to construct a geographic interface for the
discovery of the reference resources used by researchers studying historic
events. Ephemeral Cities (LG-30-03-0285-03) was something of a digital
library Oz. It may yet be something of a dream.

As 2008 opens, the dream is alive. O'Reilly plans two simultaneous
conferences: one on ETech and its influence on society
(http://en.oreilly.com/et2008/) and another on Graphing Social Patterns
utilizing GPS technologies (http://qraphingsocial.com/). Google and a
much larger "Metaverse" community (http://metaverseroadmap.orq/)
moves ever further along its roadmap toward full integration of the
participatory web with the real world. And, at the same time, Microsoft
Live Labs' Photosynth (http://www.labs.live.com/photosynth/) shows us
how the past might have looked through the aid of its artifacts.
Simultaneously, GSMA, the global trade association of mobile phone
operators, meets in Barcelona (http://www.mobileworldcongress.com/) to
advance a range of 3G devices that should continue to evolve toward a
Tech or Library 3.0.

What we aspired to do was not wholly new, just as 2007 and 2008's
technologies will have come from some predecessor concepts precepts,
if you will. We, it was decided, would build on the experience of the
Alexandria Digital Library (www.alexandria.ucsb.edu) at the University of
California Santa Barbara and the Perseus Digital Library
(http://www.perseus.tufts.edu/) at Tufts University. But, we would use the
foot-print of maps much more pro-actively than did or yet does either the
Alexandria or the Perseus digital libraries And, while we would mine our
digital library's content deep into its bibliographic core as the Perseus
Digital Library had shown us, we would bring our cores back to the surface
map. Little did we know how deep we might have to go, or, how much like
runes the exposed cores would come to seem.


final project report page 3

In our tasks, our goal seamed simple. We would create an historic atlas of
Florida using the Sanborn Fire Insurance Company1 maps, detailed for
our target cities: Gainesville, Key West, and Tampa. We would link
historic resources to that atlas, providing new geo-temporal methods of
discovery. And, we would construct learning modules that would acquaint
K-12 students with these methods. While the stuff of digital libraries: the
collection and digitization of content was solid, the technology on which
we built was something more like sand. This is not to say that the
technology could not support the structures that we endeavored to build.
But, it should be said both that these sands required unanticipated
shoring-up and that technologies that they may aspire to become is not -
indeed, no where near sufficiently stable. In retrospect, even as a proof
of concept, the breadth of the project, as ascribed by the full albeit
historically limited geography of the three cities, was inappropriately
broad: physically and temporally.

This report documents of our findings. While they characterize our
failures, they outline too the foundations for future indeed, on-going -
effort to enable the geo-temporal map.

But, why continue?

Effort associated with the Metaverse Roadmap, similar and secondary
projects, lead us head-long into the future. Moreover, the steady advance
of 3G devices that now not only phone home, but do so with exact Earth
coordinates, still and motion pictures, and other data, as well as audio,
compel us forward. Early examples of the device-enabled "geotagging"
technology are now demonstrated in Flickr (http://www.flickr.com/map), on
Panoramio (http://www.panoramio.com/map/), and by SmugMug
(http://maps.smuqmuq.com/), as well as in Google Earth and its KML
Gallery (http://earth.qooqle.com/qallery/) and Google Maps
(http://maps.qooqle.com/). Most of these demonstrate photographic
archives uses. The Google applications demonstrate the utility of
embedded data and references in addition.

1 Sanborn is the registered trade-mark of the Sanborn Fire Insurance Company, which is
wholly owned by Environmental Data Research. Neither the Sanborn Fire Insurance
Company Maps of Florida (a UFDC and PALMM digital collection) or Ephemeral Cities
are in any way connected to or supported by the Sanborn Fire Insurance Company or
by Environmental Data Research.


final project report page 4

It used to be that archivists and historians acquired artifacts and coaxed
relational data and relevance from them as so many pieces of a puzzle.
These same artifacts, created with the aid of new technologies, suggest
three-dimensional chess, leaving the researcher to see through time and
space as never before. The artifacts that we collect and interpret in the
future will come to us informed. They will have been molded, caused in
part in the context of their accompanying data. As not to waste this wealth
of information, we must create constructs that will enable us to visualize
the past, to step back into it, not as we envision the past to have been but
as suggested by the data at hand.

The geo-temporal map is one such construct. The contextual data
produced by a 3G device, while more coherent and more complete, is
similar to the contextual data that archivists and historians now produce to
understand events. It's a law of corporeal existence that everything,
whether we know or record it precisely, exist in space and time. One
wonders at the relative importance of a technology such as Twitter
(http://twitter.com/). And, we probably don't care to dwell long on it here.
Twitter allows its users to exchange messages no more than 120
characters a very simple sentence in answer to "one simple question:
What are you doing?" Certainly, the logs of the Twittering masses appear
filled with inane purpose. But, context engenders understanding of
meaning and importance. Twitter is a Tech 2.0 technology, born in 2006.
What it if been born in 2001 and had captured the character of the nation
on September 11? And, what if, rather than logged on pre-GSM mobile
phones, 3/3.5G (GSM, UMTS, or HSPDA) devices allowed location data,
images, and other vital information to have piggy-backed the logs?
Research uncovers this data, whether related to Twitter logs or historic
letters. It digs deep into news accounts, correspondents and their lives,
looking for contextual or corresponding data, mining them like Greenland
ice-cores in hopes of recreating a past climate, which is to say social
engagement and its character.

We build rules of social engagement, as we always have with each new or
successive technology. Arguably to some extent, we succeed on the
bases of input and instinct: what we know and what we think we can do
with it. It's as though we're riding along in H.G. Wells' Time Machine, with
two seemingly distinct paths into that future. In one, very much like the
2002 movie remake of the novel's original 1960 adaptation, we have the
ability to go back and re-make history. And, why not; history is so
entwined with recorded actions that it might be action and social


final project report page 5

interaction itself. Perhaps, this truly is the future of a future participatory
web. It will have an unlimited number of Hamurabi [sic]2 re-plays until we
get a desired outcome. In the other future, more like the Time Machine's
1960 adaptation, we have the ability to go back and pull life-changing
information forward with us. We never change the past. But, we use it to
change the future. So less romantic; and, certainly less actionable isn't
it? Of course, with the current state of time travel, it might be wiser of us
to suggest that information is action. Information is history. And, this,
rather than argument over the utility and value of enabling technologies, is
why we continue to encourage the construction of the geo-temporal map.

This isn't a quarrel with the gaming culture or other cultures at the fore of
the developing participatory web. Neither of these futures wants for
acceptance of the notion that information is action. Whether information
flies off the leaves of a page or out of the muzzle of a (computer
generated) rifle is ultimately less consequential than understanding the
information. A bullet is to a war game what a bullet is in the journals of 28
June 19143 or 22 November 19634 or even yesterday's TV-news crime
report. We continue to introduce more information into the gaming culture.
We want the future to re-act with awareness of the past, of the constraints
of the cultures from which we've come. The future always converges.5

As a practical mater, continuation is less about philosophies than about
being suited to research. Our proposal's narrative worded the argument
this way: From Gainesville to Key West, a sense of place gives meaning to
our lives. Your great grandfather bought a citrus grove in Eustis; mine
worked in the pencil factory near Way Key now Cedar Key. Where we're
from, where we're going all revolves aroundplaces with names. The social
fabric of modern cities reflects the rich mosaic of activities of past

2 Hamurabi: an early computer game. The fate of Hammurabi's kingdom rest is his ability
to manage commodities. Most player perfect their technique and accuracy relative to the
historic record by playing the game over.
3 28 June 1914 marks the start of the first World War: an assassin's bullet kills the Grand
Duke of the Austro-Hungarian Empire in Sarajevo, Bosnia-Herzegovina. It was, as it was
for my grandparents reading the news in Vienna, an "everyone-remembers-where-they-
were" moment.
4 22 November 1963 marks the assassination of John F. Kennedy, 35th President of the
United States of America. It was, as it was for my mother her soap-opera interrupted
with the news another of those "everyone-remembers-where-they-were" moments.
5 This has to be a better way of acknowledging George Santayana's Life and Reason :
Volume 1, Reason in Common Sense (1905) than does continuing to paraphrase him
more than a century later: those who cannot remember the past are doomed to repeat it.


final project report page 6

inhabitants. Historically, place identity and place attachment have been
associated with the development of peoples' attitudes, values, and beliefs.
We believe that these associations, these motivations for action will
continue long into the future. We endeavor to create a construct for better
understanding. And, despite for failures of the Ephemeral Cities project,
the existence of new web-applications such as Flickr, Google Earth,
Panoramio, and the like justify our vision. Work remains.



The Digital Library Center at the University of Florida's George A.
Smathers Libraries (hereafter, UF-DLC), took the project lead. It
maintained responsibility for budgets, production targets, and digital library
systems and other programming, as well as partnerships. For tasks
involving the creation of geographic information systems (GIS) layers, the
UF-DLC was joined by the Libraries' GIS Program in the Government
Documents department.

Additionally, the UF-DLC took on the role of local lead for the City of
Gainesville, Florida. We coordinated a local partnership including the
Alachua County Library District, the Alachua County Historic
Trust/Matheson Museum, and the Alachua County Clerk of the Court.
And, through the auspices of the Library District and the Historic Trust, we
were joined in the provision of artifactual content by the citizens of
Gainesville and of Alachua, Levy and Marion Counties.

Other local leads were taken by the Digital Library Center of the Florida
International University (hereafter, FIU-DLC) and the Digital Special
Collections of the University of South Florida's Libraries (hereafter, USF-

FIU-DLC took on the role of local lead for the City of Key West. Its chief
officer, Gail Clement, coordinated a local partnership including the May
Hill Russell Public Library of Monroe County Public Library, the Key West
Art & Historical Society, and the City of Key West in collaboration with its
Public Records Office. And, through the auspices of the Public Library
and Art & Historical Society, the FIU-DLC group was joined in the


final project report page 7

provision of artifactual content by the citizens of Key West and Monroe

USF-DSC took on the role of local lead for the City of Tampa and, in
particular, its Ybor City neighborhood. Its chief officer, Dr. Mark
Greenberg, coordinated a local partnership including the Tampa Bay
History Center and the Henry B. Plant Museum. The USF-DSC was
joined in the provision of artifactual content by the citizens of Tampa and
Hillsbourgh County.

We were joined, for the provision of digital library services and digital
archiving by the Florida Center for Library Automation (FCLA), based in
Gainesville, Florida. The Florida Center for Instructional Technology
(FCIT), based in Tampa, Florida, was to have joined us for the provision of
learning modules. Delays in our programming schedules, however,
resulted in FCIT's withdrawal from the project. Learning module design
was assumed by the UF-DLC.

The Cities.

Gainesville, Key West and Tampa were chosen for the Ephemeral Cities
project as much as for the willingness and strength of local partnerships
as for the content and historic significance of the three cities.

Gainesville was an important rail-town. During the Seminole Wars, the rail
lines through Gainesville marked the southern extent of contiguous settled
Florida. Centrally located in historic Florida, Gainesville was a processing
point for lumber harvested from north Florida's pine stands. Sanborn
Fire Insurance Company maps of Florida's largest cities Jacksonville and
Tampa showed evidence of Gainesville's lumber before 1906 when fire
claimed Jacksonville whole and prosperity rebuilt Tampa brick-over-
mortar. More importantly, Gainesville was also a collection point for
agricultural harvests on their way to market along the U.S. Atlantic
seaboard. Produce was destined for New York and New England where,
during the period 1880-1920, the project target, Florida tourism promotion
was booming. Railways, that carried produce north, carried tourists south,
or, as one of Jacksonville's economic news journal giddily proclaimed,
"Tourists in! Vegetables out!" The headline has been long-since true of
Florida, and Gainesville at one time provided the grist for its industrial


final project report page 8

While vegetables flooded out of Gainesville, the tourists flooded toward
Miami and Key West. To this point, Miami had been a beach-front
community for the hardiest of settlers. A golden spike had been driven
into the Earth in the wilds of Utah long before Miami and Key West would
be connected by rail. Until 1912, this part of Florida was still frontier! Key
West had long since been founded however. It's origins rest in the days
when ships drove the commerce of the Americas. Key West was the
United States' ambassador to the Caribbean, and, its defense from the
colonial powers that plied its waters. Key West was both trading post and
military out-post. It was arguably the most international of the three target
cities. The name Key West derives from the city's original Spanish name,
Cayo Hueso. In more modern times, Key West served as President
Truman's Winter White House, a bridge to Cuba just ninety miles away,
and the home of great American writers Ernest Hemmingway and
Tennessee Williams among them. The Key West that preceded them,
from 1880 through 1920, had been more Caribbean than American, and
more island nation Conch Republic than seat of Florida's southern-
most county.

As rail lines pressed south-west from Jacksonville and Saint Augustine
through Gainesville, they arrived at Tampa. Until then, it had been
connected to Florida and the nations by sea routes. Tampa was a sea
port for the out-flowing riches of Florida's natural bounty. Shipping lanes
rounded the peninsula of Florida through Key West and made landfall at
East Coast ports from Jacksonville northward. Gulf Coast ports, notably
New Orleans, provided more direct trade. Their voracious appetites fed
heartily and particularly on Florida's industrial exports notably phosphate
and lumber during the Reconstruction and the Great Expansion that
followed. The timbers that Tampa built upon prior to the railroad's arrival
had also lumbered in the making of New Orleans. Indirectly, the railroads
changed the character of Tampa. The change can be seen clearly on the
Sanborn Fire Insurance Company maps, as brick replaced wooden
structures between 1880 and 1920. While some cities like Jacksonville
would be made-over almost wholly by fire, prosperity more so than fire
insurance made Tampa anew. St. Augustine and Augusta brick were the
currency that paved Tampa's streets and built its large cigar houses. The
opening of Sanchez and Haya cigar makers in 1886 followed the opening
of the rail lines by less than a year. The cigar industry would peak in
1929. So, the period of 1880 through 1920 marked the industry's heyday.


final project report page 9

Together, these three cities Gainesville, Key West and Tampa as seen
from 1880 through 1920 document a state during its transformation from
frontier to precursor of today's contemporary Florida. Unlike the other,
more established Florida cities of the day, these three stood on the
frontier. These cities were Florida's Khartoum, Cape Town and Dakar,
encircling a then unknown territory on the cusp of boom and bust.


Each of the three local partnerships met their digitization targets.
(Additional discussion of targets follows.) And, each of the three local
partnerships, building on past projects, was strong. The Key West
partnership faced the challenge of geography more so than other local
partnerships. FIU-DLC is approximately 100 miles from Key West. A
previous relationship, growing from the Monroe County Public Library's
LSTA-funded Mile Markers (http://palmm.fcla.edu/mile/) project prepared
them to complete this project.

Each of the local partnerships grew and maintained a slightly distinct local
style. And, it appears that each partnership continues in a variety of ways.
All of the major and most of the minor partners continue to digitize local
content that feeds into Ephemeral Cities discovery systems. Not all,
however, continue to enable their content with the project's geo-temporal
tagging methods. Time and labor costs are major factors. But, a
transparent (i.e., widely available) and mature database functioning as an
authority and seamless query enhancement system is also a factor.
(Additional discussion of authority systems follows, see Appendix B.)

Florida, itself, proved greatest challenge to each local partnership. Three
hurricanes of 2004 took aim at us, and, some of us suffered blows from
more than one. It seemed that nature would have us be the more
ephemeral. It was inevitable that one city-portrait photography trip would
be nearly cancelled. But, Key West blew off the storm, and, the advanced
departure of tourists left Key West, in a way, pristine for architectural
photography to round out the collection of texts and artifacts. Gainesville
and particularly Tampa weren't as fortunate. Tampa suffered the nearly
head-on blows of two hurricanes. And, Gainesville, though it suffered the
glancing blows of only two hurricanes, felt them with peculiar weight as
storms of such magnitude rarely follow an inland path. The damage
inflicted by old trees upon old houses though it gave us opportunity to


final project report page 10

explore old infrastructure also left us, and particularly our staff, to deal
with personal recovery issues.

PALM Ms Grow in Florida.

It was a storm of more human dimensions, however, that presented the
greatest challenge to the project. Ephemeral Cities was to have been
another of the PALMM Collections (http://palmm.fcla.edu/). Its three
principals, the University of Florida, Florida International and the University
of South Florida, were all major contributors to PALMM, the collaborative
digital library of Florida's universities. Each agreed to make the new
content funded by the project available in PALMM. And, the Florida
Center for Library Automation, which agreed to provide hosting and digital
archiving services, was responsible for the daily operations of PALMM,
was also to have become responsible for maintenance of Ephemeral
Cities' digital content. But, a variety of circumstances, both technical and
human, resulted in deviation from the plan.

To be certain, each of the partnerships maintained and continues to
maintain its digital archiving responsibilities. The IMLS-funded Florida
Digital Archive (http://www.fcla.edu/digitalArchive/) is a model for digital
archiving. With continued funding from the State of Florida, the Florida
Center for Library Automation continues to facilitate the service. And,
each partnership takes advantage. Resources generated through the
auspices of Ephemeral Cities are and continue to be retained and cared
for in the Florida Digital Archive. UF-DLC, in agreement with the Alachua
County Historic Trust, takes responsibility for the Gainesville component
project's materials. FIU-DLC, in agreement with the Monroe County
Public Library, takes responsibility for resources generated for the Key
West component project's materials. And, USF-DSC, in agreement with
the Plant Museum, takes responsibility for resources generated for the
Tampa component project's materials. And, resources created centrally,
e.g., digital city directories, city newspapers and the Sanborn Fire
Insurance Company public domain maps of Florida, are archived under
the authorization of the UF-DLC.

Among the component projects, the Key West partnership remains truest
to the original hosting plans. FIU-DLC has a vibrant relationship with the
Florida Center for Library Automation. Like the Mile Markers project
before Ephemeral Cities, its Ephemeral Cities resources have been


final project report page 11

contributed directly to the PALMM Collections, notably its Florida Heritage
Collection (http://palmm.fcla.edu/fh/). Some of its content has also been
duplicated in the Mile Markers collection, hosted by the Monroe County
(Florida) Public Library (http://www.mile-markers.org/).

The Tampa partnership has released selected content to the PALMM
Collections, notably its Florida Heritage Collection, but also to other digital
libraries. Many, if not the bulk of Tampa resources, have been posted to
Floridiana on the Web (http://www.lib.usf.edu/ldsu/digitalcollections/
F03/html/). Floridiana on the Web is both a PALMM Collection and a
component of the USF Libraries Digital Collections. In fact, both are
specialized interfaces of USF-DSC digital asset management systems
(DAMS). Because USF-DSC was transitioning between DAMS during the
project, much of its Ephemeral Cities content was also posted to the UF
Digital Collections (UFDC) (http://www.uflib.ufl.edu/ufdc/). USF-DSC's
DAMS is now run under the ExLibris DigiTool application

The Gainesville partnership has released the bulk of its content to the UF
Digital Collections (UFDC) (http://www.uflib.ufl.edu/ufdc/). But, because
the Gainesville partnership was the most complex its independent
partners hosted their own digital collections, Gainesville's Ephemeral
Cities content was also mounted in sub-component parts to the Alachua
County Clerk of the Court's Ancient County Archives (a.k.a., Ancient
Records) site (http://www.clerk-alachua-fl.org/archive/), to Alachua County
Independent Genealogy (http://www.rootsweb.com/-flalachu/), and to the
Alachua County Library District's Heritage Collection
(http://heritage.acld.lib.fl.us/). The Alachua County Historic Trust's
Matheson Museum maintains a special relationship with the UF-DLC. Its
Ephemeral Cities content are posted to UFDC but retained as separate
collections, e.g., Matheson Museum Visual Resources Collection
(http://www.uflib.ufl.edu/ufdc/?s=mhcc) and Matheson Museum Oral
History Collections (http://web.uflib.ufl.edu/ufdc/?c=mmoh). All Matheson
content is prepared for use by Ephemeral Cities applications. Gainesville
content originating with the University of Florida and from the Matheson
Museum is selectively duplicated in the PALMM Collections, notably in its
Florida Heritage Collection. The Sanborn Fire Insurance Maps of
Florida, for example, used to form the base layers for the Ephemeral
Cities project, is available in either a PALMM interface
?c=Sanborn&n=ralmm) or a UFDC interface


final project report page 12

?c=Sanborn), together with the Ephemeral Cities project extensions

Other digital content, available for Ephemeral Cities search, but created
outside the project is found in digital collections throughout the state. The
majority of these collections external to the project are harvested by the
Florida Center for Library Automation in Florida Electronic Library's Florida
on Florida collection (http://purl.fcla.edu/fcla/floridaonflorida). The State
Library and Archives of Florida's Florida Memory Project
(http://www.floridamemory.com/) holds the bulk of external resources
documenting Gainesville, Key West and Tampa, as well as virtually all
other future targets of the Ephemeral Cities project.

Why Coconuts Fall From the Tree.

The PALMM tree parented these somewhat dissonant collections for both
technical and human reasons. PALMM is a loose confederation of digital
projects, many collaborative, some dedicated to the work of a single
institution. The PALMM Collections, for the most part, are topically
Florida. Most PALMM Collections avail themselves of the centralized
services and digital library technologies of the Florida Center for Library
Automation (FCLA). At the time of Ephemeral Cities launch, PALMM
Collections steadfastly observed a common look and feel, whether they
utilized the centralized services of FCLA or the distributed services of the
constituent institutions. PALMM institutions ranged in size from just over
3,500 students to well over 45,000 students. And, the knowledge, skills
and experience of both their faculty using digital collection and their digital
collections staff building digital collections varied as widely, even among
the principal partner institutions. At the time of Ephemeral Cities launch,
the centralized services were bound by the abilities and limitations of the
DLXS digital library software. These factors, not fully appreciated during
Ephemeral Cities' planning, bore the seeds of change.

Among the human causes for change, one issue stood foremost as
problematic: branding. Materials contributed to PALMM Collections,
retained attribution of ownership and contribution in metadata in citation
and bibliographic data, but were otherwise outwardly marked with the
PALMM brand. Institutions wishing to leverage their work for additional or
subsequent funding found themselves having to explain the apparent lack


final project report page 13

of institutional branding. The issue was so problematic for one curator of
target materials at one of the principal partner institutions that the curator
fought to withhold the needed content. When PALMM content was
harvested, as it was for AmericanSouth.org (http://AmericanSouth.orq)
and the National Science Digital Library (http://nsdl.orq/), it bore the
attribution of the Florida Center for Library Automation rather than that of
the contributing institution. This issue was so problematic for one curator
of target materials at another of the principal partner institutions that the
curator's institution insisted on building first under the institution's aegis.

Among the technical causes for change, DLXS software or, more
specifically, our ability to program for it became problematic. DLXS was
somewhat brittle or, at least, insufficiently flexible. It wasn't bad for what it
was. And, the Florida Center for Library Automation (FCLA) made it do
things not originally within the functional scope of the application. But, its
ability to mine and use mined data was less optimal than required or than
planning estimates gauged.

Planning assumptions about how the project would come together were
being set aside by time funding enabled work. FIU-DLC continues to host
content solely with PALMM. FCLA, however, is now migrating PALMM
collections and content into the Ex Libris DigiTool application, with plans
for future migration of selected content to a digital library rolled out under
Fedora (http://www.fedora.info/). USF-DSC now hosts content under its
own instance of the Ex Libris DigiTool application, with some of its content
duplicated PALMM. UF-DLC and the Ephemeral Cities site is now hosted
under a suite of services collectively known as the University of Florida
Digital Collections (http://www.uflib.ufl.edu/UFDC/technical/index.htm). Its
data stores are maintained under MS-SQL. Its digital library store, under
Greenstone (v.2) currently; and there are plans for a future Fedora
migration for additional control and more robust structures. Component
digital objects run under separate services. Scalable images, for example,
are maintained on servers currently running the commercial Aware
JPEG2000 Server application; and, migration to an open-source .NET
JPEG2000 solution is planned for increased control. Digital audio and
video run under yet another server. University of Florida Digital
Collections (UFDC) also maintains an independent and locally
programmed query and presentation layer. The power Ephemeral Cities
rests largely here.


final project report page 14

A continuing problem for Ephemeral Cities is consistency of geo-temporal
tagging. The DLXS and DigiTool applications do not fully support, either
or both in collection or query, the extent of data required by Ephemeral
Cities. Our ability to harvest the needed data is limited. And the ability of
the partnership to provide it, that is to continue entering it into bibliographic
records or metadata, is taxed by limited funds (compounded by the
inability to demonstrate cost-justifying utility at institutions using DLXS and

Our Objectives

Ephemeral Cities laid out five objectives:
1. Build a Geographic Information Systems (GIS) Interface.
2. Digitize relevant historic resources.
3. Invite local, general participation in and contribution to Ephemeral Cities.
4. Create education modules building research skills.
5. Promote the model.

Ephemeral Cities' web presence was constructed as part of the University
of Florida Digital Collections (http://www.uflib.ufl.edu/ufdc/). Content
supporting the project, insofar as contributed to Ephemeral Cities
centralized servers,6 was deployed as topically appropriate across those
collections. The main project interface, Ephemeral Cities proper, was
deployed with a virtual "subcollection" interface (http://www.uflib.ufl.edul
epc/). The interface supported a variety of search methods: selection
map, basic and advanced textual searches, and browses by city (Fig. 1).

6 There was no requirement for contribution to the project's central servers. Content could
be deposited to those servers or could reside in any of the recognized partnered-
institutions' digital libraries. Contribution to the central servers was highly recommended.
When content was deployed to a local/remote digital library, contribution of metadata and
text was encouraged. Barring that, metadata harvesting was attempted, although not
always successfully for reasons describe elsewhere in this report. In retrospect, a project
this complex should have required contribution to the central servers (this was
complicated by the political discourse of content deployment also discussed elsewhere in
this report).


final project report page 15

r,,Fa emeral Cities Project


Search: Map I Basic I Advanced
Browse by City: Gainesvlle I KeyWest I Tampa

Figure 1. Interface search options.

Search by SELECTION MAP (Fig. 2a) looks like a tool with which many of
our patrons are already familiar, that of the Florida Electronic Federal
Depository Library (FEFDL)7 Patrons of the Ephemeral Cities site select
any of Florida's counties by clicking the county outline on the map.
Mouse-over actions identify counties by name. Currently, Ephemeral
Cities target cities are Gainesville, in Alachua County in central northeast
Florida; Tampa, in Hillsborough County on central Florida's Gulf coast;
and Key West, in Monroe County at the southern-most part of Florida. In
anticipation of Ephemeral Cities' growth to other cities in other counties of
Florida, this part of the interface links the user to available public-domain
Sanborn Fire Insurance Company Maps of Florida.8

Selection of a given county from the map links to a county information
page under the Ephemeral Cities banner, which lists the public-domain
Sanborn map sets available for the county (Fig. 2b). The information on

7 Florida Electronic Federal Depository Library (FEFDL) http://www.uflib.ufl.edu/fefdl/ was
funded in part by Florida's LSTA. The map interface serves as a gateway to government
8 In anticipation of Ephemeral Cities' growth, each of the 3,167 sheets in the Sanborn Fire
Insurance Company Maps of Florida digital collection have been geo-rectified. Not all
have been made accessible, however, within the Ephemeral Cities GIS interface. Digital
objects entering at least UFDC collections, and many entering the PALMM collections,
have geographic referencing of one sort or another. We envision that some Ephemeral
Cities or Ephemeral Cities like interfaces will develop on top of services outside of UFDC.
The Central Florida Memory digital library, which maintains an extensive collection of
central Florida content based on the technologies deployed at the University of Central
Florida, is a likely candidate for such interfaces covering Orange and surrounding
counties. The Digital Library of the Caribbean, a project based on UFDC but based at
Florida International University (one of the Ephemeral Cities partners), has discussed a
not-dissimilar map interface for the Caribbean. Should that interface be developed, it
would likely reside at either the University of Florida and in UFDC or at Florida
International University on servers currently running the Andean Amazon Geographic
Information Systems (GIS) Web Portal (AAGWP http://aaqwp.fiu.edu/), another IMLS
funded project.


final project report page 16

this information page may be sorted by city name or by year of
representation, either ascending or descending. Selection of a set links to
a visual index (Fig. 2c) for the city and date selected.

,,Eaemeral Cities Project -...L. ..

Figure 2a. Selection map.

Ephemeral Cities planners had actually proposed a more seamless
experience like that now experienced with Google Maps
(http://maps.qooqle.com/), Google Earth (http://earth.google.com/), or the
United States Geologic Service's "National Map Seamless Server"
(http://seamless.usgs.qov/), none of which were available at the time that
Ephemeral Cities was planned. The ArclMS software to which the project
was tied9 was incapable of supporting the seamless experience.

9 The project was tied to ArclMS as a matter of cost. ArclMS came to us virtually free of
cost via a campus license. Project planners had inadequate information regarding
ArclMS limitations and were not disabused of our hopeful notions. Replacement with
another commercial service or construction of a wholly new service would have been
impossible in any case. The project budget and project time-table, both, were already at
the maximum allowed by IMLS. Now fully aware of ArclMS' limitations, the project
planners, programmers and GIS librarians have begun to discuss alternatives. At this


Search: Map I Basic I Advanced
Browse by City: Gainesvlle I Key West I Tampa


4 4 ,... I ..



final project report

.,Ephemeral Cities Project .. .........


Search: Map I Basic I Advanced
Browse by City: Gainesville I Key West I Tampa

Monroe County
I- :. I 3t.I :I-. ] 1 : _: .
I-I r,.. ,.]., r r,,- ] n ,- J ,,' n j3 ,,- r lj.,, i U ; 1r,1 1i:j
Major Cities: Islamorda, Key Largo, Key West, Marathon, North Key Largo Beach

Th-ciL .LrF \I rl.p 4 i L :,i F:,r r,,-: ari L ':,Frt tv ,r
Key West YearAscending
Key West, Monroe County, Florida, 1889 [visual index I full citation ] Year Descending
Key West, Monroe County, Florida, 1892 [visual index I full citation ]
Key West, Monroe County, Florida, 1899 [visual index I full citation ]
Key West, Monroe County, Florida, 1912 [vsual index I full citation ]

Figure 2b. Selection map: County selection

The visual index (Fig. 2c) in turn links the patron to map sheet displays
(Fig. 2d) representing individual city sections for the specific date of the
parent visual index. Sheet displays are available in standard view, a
JPEG image measuring 630 pixels wide, or in zoom view, JPEG images
"parsed" from a JPEG2000 image, which can be manipulated to various
sizes and scales.10 An option to view the sheet or a portion of the sheet in
the GIS interface (see below) should also be available.

time we are unable to recommend any alternative, though some alternatives are viable
and promising. One such alternative, currently not without its own set of limitations, is
discussed below.
10 Zoom image behaviors are those standard of the Aware, Inc. software for digital archives
(see, http://www.aware.com/imaging/digitalarchives.htm). Project planners are
considering a move away from Aware toward a solution that supports pan as well as
zoom. We are also considering a solution that supports Regions of Interest (ROI), part of
the JPEG2000 standard, supported by Aware but which we are not now using, or a
solution using Asynchronous JavaScript and XML (AJAX) and vector map engine. ROI
and XML solutions would populate the map with data from metadata and text files and
from the SQL tables currently used to store index data and digital object references. We
are currently studying options and not prepared to move from the current, albeit
unsatisfactory, solution.

page 17

t b-*- --


final project report

trearshni Gollibe
larrh All Molasins.
Rtal'd Map a:s
F.ll CHL:.

All Map Itees


Map and Imagery
I ri r y


Title: Key Wet. More County. Florida. 1B9

Visual Index

Figure 2c. City/Time-Period display

The BASIC SEARCH (Fig. 3a) provides the patron with a single, Google-like
keyword query box. Queries search the Sanborn Fire Insurance
Company Maps of Florida digital collection's index of county, city and
street names and of features both natural and man-made. This part of the
interface is intent on finding place. Query does not search the digital
collections at this point.

The original plan for Ephemeral Cities would have allowed the patron to
research the geo-temporal continuum from either the map interface or the
textual interface. The current execution of Ephemeral Cities is strictly a
map interface, intent on finding first place, limiting discovery to temporal


page 18

t b_* --

^BB JLjjI J -I i .1,JLJLY^A JiJL-L i- tlt~k-r^uL&

final project report

zone, and only then searching text within the zone. While the interface
aids the researcher, it presently forces the researcher to navigate through
too many steps.


F l11 CAllab
Fea Erels

1 r-
All Map 1SK ,




* ,. ~ .
-" -" n- -
Kuf '


- ..'. -. *

i'-" -- ".


Figure 2d. City section display (standard map also available in zoomable version)

It is possible for the researcher to search textual collections and metadata
regardless collection and then to reveal geographic and temporal data in
the "full citation". The method is not transparent. And, because the data,
here, is not yet fully linked back into the map interface, the researcher
must launch a separate query in that interface. This again forces the
researcher through hoops. This problem should be fixed through future
enhancements to the project applications.

page 19

Title: Key West, MomoeCounty, Florida, 88

|F i,9F- -s-I'\ GDato:l 3n Fi4 a -1-1

S "-: "-

: -- 4.

S.. .


Map and Ima gry
L i b r y


FI l l r -







final project report page 20

Fbemeral Cities Project A ... ..


Search: Map I Basic I Advanced
Browse by City: Gainesville I Key West I Tampa

li:yW ,:,Fr:j0:; | 5:. n:r,,:::ll:::,ii r, |
Searches namrnes f ties. muntles, streets. mmpanles, and building uses Me a3sut bDasiLearch

Figure 3a. Basic textual/index search

general Cities Project -


Searcth Map I Basic I Advanced
BrowsebyCity Gainesville I KeyWest I Tampa

Keyword(s) hot el Search Collection
Search& r~n elsa oxrmnTi, s reet, ocrnr aMr IiL g Ld Ulr kre aWb.: tasc search

There were 575 matches on building uses with your keywords
Orderresultsfirstby Buliding Use v ,thenby Company [ ChangeOrder

Acme Hotel
no company listed
Jacksonville, Duval County, Florida, 1891. [sheet 11 I visual index I full citation 1
Jacksonville, Duval County, Florida, 1897. [sheet 51 visual index i full citation
Jacksonville, Duval County, Florida, 1897. [ sheet 51 visual index I full citation]
Alabama Hotel, The, Boarding
no company listed
Plant City, Hillslorough County, Florida, 1914. [sheet 6 [visual index [ full citation]
St. Petersburg, Pinellas County, Florida, 1913. [sheet 19 visual index [ full citation]
Alachua Hotel
no company listed
Gainesvile, Alachua County, Florida, 1&87 [sheet 3 [visual index full ctaton ]
Figure 3b. Query results


final project report page 21

The Basic search, like selection map searching, returns data related to the
Sanborn Fire Insurance Company Maps of Florida sheets (Fig. 3b). The
drill down to data is very much the same. Unlike the selection map, Basic
search drills down into the GIS interface, where the digital collections will
be searched.

The ADVANCED SEARCH (Fig. 4a) provides the patron with the ability to
differentiate a search. Searches may be limited to building use,
commercial name, etc. and further limited to a specific city for a specific
date range. Search results (Fig. 4b) may be sorted by building use and
company name.

,.Enhemeral Cities Project ..._ ...


Search: Map I Basic I Advanced
Browse by City: Gainesville I Key West I Tampa

Finr',l .' | I', : W n [Ir, !.. : W ',:,rr:lT I:

that appear on maps of AnyCity v
between the years of 11 v and 1924 Search Collection

Figure 4a. Advanced textual/index search

Again, queries search the Sanborn Fire Insurance Company Maps of
Florida digital collection's index and remain intent on finding place. Query
does not search the digital collections at this point. Unlike the selection
map, Basic search drills down into the GIS interface. Here, the digital
collections will be searched.


final project report

Figure 4b. Query results

BROWSE BY CITY (Fig. 5a) directs the researcher to the three Ephemeral
Cities currently targeted: Gainesville, Key West and Tampa, and provides
a selection of years falling within the project window. Though additional
year sets all geo-rectified exist for each city, only those year sets
originally targeted during project planning are available in the year set.

Selection of a given year links the researcher to the visual index for the
selected city/year pairing (Fig. 5b). The visual index, as seen in the
selection map and basic and advanced searches, is a JPEG image
presented with an HTML image-map. Again, selection of a particular city
section links to a view of that city section. At this point, however, the link
is directed into the GIS interface (Fig. 5c).

page 22

h Ehermeral Cities Project j ..


Searcf Map I Basic I Advanced
BrowsebyCity Gainesville I Key West I Tampa

Find a Building Use v worth the keyword(s) hotel

that appear on maps of Any City V

between the yearsof 184 and 1924 v SearchCollection

There were575 matches on building uses with your keywords,
Orderresultsfirstby BuilingUse vthenby Company .i Change order

Acme Hotel
no company listed
Jacksonville, Duval County, Florida. 1891. [sheet 11 I visual index I full citation I
Jacksonville, Duval County, Florida, 1897. [sheet 51 visual index [ full citation ]
Jacksonville, Duval County, Florida. 1897. [sheet 51 visual index [full citation ]
Alabama Hotel, The, Boarding
no company listed
Plant City, Hillsborough County, Florida, 1914. [sheet 6 visual index full citation]
St. Petersburg, Pinellas County, Florida, 1913. [sheet 191visual index) full citation]
Alachua Hotel
no company listed
Gainesville, Alachua County, Florida, 1W87. Sheet 3 visual index] full citation]

final project report page 23

,FEjaemeral Cities Project g-t .L. .. ..


Search: Map I Basic I Advanced
Browse by City: Gainesvlle I Key West I Tampa

Key West

Choose your year of interest:
Figure 5a. Browse by city

interface can be difficult for a first time user and, indeed, for many return
visitors. The Ephemeral Cities GIS Interface is a simplified interface.
Buttons and functions not needed by the average researcher have been
removed. Tools and on screen clutter, kept to a minimum. The ArclMS
application underpinning the Ephemeral Cities GIS is relatively inflexible in
the use of screen real-estate. Further complicating the interface are the
limitations of the present implementation and the limitations of the public
schools and home users that we target. That the present interface is
designed using 194 city section maps limits utility through added
complexity; discussion follows. That the public schools and most home
users are one to two generations behind the university-computer
environment supporting Ephemeral Cities further limits screen real-estate
and the amount of information that can be delivered reliably.

Four Ephemeral Cities education modules provide user instruction.
Navigating Ephemeral Cities, Part One: Alternate Routes for
Finding Information (
edmod1 files/UF/01/03/01/81/00001/UsingEphemeralCities. pdf)
Navigating Ephemeral Cities, Part Two: Alternate Routes for
Finding Information (
edmodi files/UF/01/03/01/82/00001/UsingqEphemeralCities2. pdf)

^BB^JL-jI- JL-IL -LiV.1, l ln A lwiJL-L i- tlt~k-r^uL&

final project report


Se-arch: Map | Basi I Advanced
Browse byCity: Gainesville I KeyWest I Tampa

Key West 1889

Figure 5b. City display by year (JPEG image map used as point of entry to GIS interfaces)

* Navigating Ephemeral Cities, Part Three: Alternate Routes for
Finding Information ( files/
UF/01/03/01/83/00001/Using EphemeralCities3. pdf)
Navigating Ephemeral Cities, Part Four: Alternate Routes for
Finding Information (
edmodi files/UF/01/03/01/84/00001/UsinaEDhemeralCities4. Ddf)

page 24

^B ^Ljj-J -L- i .,J JY^ ~ ^ LJ --L i? twi&k-r^uL&

final project report

page 25

Figure 5c. City section: the GIS display (a separate, albeit not dissimilar, GIS interface for each city section)

Ill. i -, .i

' I 1 h I I

I I' 'I 11' '' 1 '

Search all Sheets
Ephemeral Cities Home Page
Figure 5d. Searching the GIS

Zoom to records)
V ~ ~: Ti ,.... i..i r SI.,
I- t *. :- ,- ji r T,- ij i i
Zoom to these records
Figure 5e. Query returns

The user enters the GIS through one of the visual index (Fig. 5b) and is
presented with one of one-hundred and ninety-four individualized maps
(Fig. 5c).11 The blue stars on the sheet represent identified points of
interest, known places, or as they are known in Google Earth speak,

11 Ephemeral Cities originally envisioned as single interface. The reasons and problems with a
variety of sheet interfaces is discussed below.

Key West, 1889 (Sheet 4) ] Q Q ['

,\ Search

UF j. G
? Search For: GetValues

FIU ". ,",
'.[* AddtoQueryString

S'Search this Sheet

S"" Search all Sheets
'I Ephemeral Cities Home Page


final project report page 26

placemarks. Clicking on any one shows building information and a query
link. Several locations can be selected using the red dotted-box tool.

The user then constructs a query (Fig. 5d), deciding to search either
building uses or street names. This implementation of ArclMS is
decidedly limiting; the original vision was of a much broader search
including persons, other named things and events. While ArclMS may
work with SQL, it prefers relatively flat tables. ArclMS was designed
before relational databases became fashionable; and, here, ArclMS is
clearly showing its age.12

Characteristic of ArclMS' unfriendly nature, the user then clicks "Get
values" to retrieve the values associated with the map for the search type.
The retrieved values are limited to the sheet displayed. This too is
decidedly limiting. The original vision was to search across all sheets.
Here, the user is forced to search all sheets of interest individually while
working within the GIS interface. To compensate somewhat, we've also
pulled all of the values into the textual search systems of the UF Digital
Collections, systems that support search of Ephemeral Cities digitized
resources. Frequently, these GIS map sheet values are also too specific.
A search of "hotels", for example, may force the user to search the generic
"hotels" then, in a separate search, individual named hotels. This feature
or fault, however, is more reflective of an insufficiently relational database
structure than of the GIS interface. Programmers have identified this
feature as needing correction.

Still the more characteristic of ArclMS unfriendliness, the user then must
indicate the obvious: add the term to the query string. Ephemeral Cities
planners had hoped to work around this and similar limitations using
screen-scrapping techniques used, for example, by UF Digital Collections
to work around the limitations of its Greenstone Digital Library data store.
ArclMS, however, proved to be unresponsive to such creative
approaches. Currently, the interface does not support Boolean logic
though ArclMS usually does support the logic. Boolean was removed
when study indicated that our users were having difficulty navigating
ArclMS' arcane methods (define search type select from listed values -

12 Again, we chose to work in ArclMS for several reasons: a campus license (such licenses
are common on university campuses, and, we originally thought this might make the
model sustainable as it was used elsewhere) and because viable alternatives were not -
and largely are not yet available (or available within the cost constraints of the public
sector). It is fair to say that the Devil we know, was the Devil, indeed.


final project report page 27

add to query select Boolean operator define search type select from
listed values add to query) for construction of the Boolean search.
Removal of the Boolean search also freed valuable screen real-estate
from ArclMS' poor dictates on the division and use of divisions of screen

With a search of the sheet, the GIS interface returns hits or in ArclMS'
parlance: records (Fig. 5e). Records are comprised of a record number,
unique for the particular query return, building use, street name, and a link
to Ephemeral Cities digital resources and digital resources in other related
collections. The GIS interface afforded by ArclMS allowed little screen
real-estate. Though Ephemeral Cities had planned to return more
information (e.g., a proprietor or owner and secondary uses most
grocery stores of this era, for example, also served as tobacco shops), this
was sufficient information for most users. This multi-use problem together
with a street-name problem13 are the result of rigidity in the underlying
ArclMS, which prefers flat to relational database structures.

Figure 5f. Query returns displayed on the map (yellow dots here surrounded by red circles)
Figure 5f. Query returns displayed on the map (yellow dots here surrounded by red circles)

13 Key West and Tampa saw occasional street name changes. In Gainesville, street names
changed to street numbers. Today, most researchers of Gainesville history no longer
remember the city's street names. Though this relational data has been saved in
authority records, they information could not be brought into the ArclMS based GIS


final project report page 28

Each query return is accompanied by the clickable text "Zoom to
recordss), which identifies the location of returned locations on the map.
As illustrated in Figure 5f, above, locations are identified by yellow dots,
which for purposes of illustration here are within red circles. From the
query results, any single location may be identified by clicking the
location's record number.14 A selection is then indicated on the map by a
red dot (Fig. 5g), illustrated here surrounded by a yellow circle for quick
identification.15 This behavior makes possible proximity studies. But,
again because of limitations of both the ArclMS and the individuation of
city segment maps, the true potential of proximity studies is not realized in
the current Ephemeral Cities GIS interface.

Figure 5g. Selection of a particular point (red dot here surrounded by a black circle)

14 Record numbers are temporary, particular to the individual query, their impermanence
has been identified as confusing to most users. The behavior is tied to the underlying
ArclMS and will not be changed until that GIS is replaced by another system.
15 Most users consult the GIS map without zooming in. This makes difficult the
differentiation of blue stars, difficult to see the footprint of the building they identify, and
difficult to locate the yellow and red dots. Seeing the yellow dots is the more difficult,
particularly on early Sanborn" maps, where yellow represents an overwhelming number
of wooden (yellow) buildings. Tests with color changes found no better combination.


final project report page 29

Clicking a link (indicated by the camera icon) from the query returns box
launches a query against the digital collections amassed by the
Ephemeral Cities project. Original plans for this search envisioned a
broadcast search one that would search all project partner digital
libraries as well as that of the State Library and Archives of Florida.
Performance issues, however, necessitated harvesting remote data into
the UF Digital Collections platform on which the bulk of the Ephemeral
Cities project was constructed. This too has held its limitations: imperfect
or incomplete harvesting; remaining current with additions to partners'
digital libraries; and harvesting's inability to harvest searchable text.16

Fjnbemeral Cities Proiect c J


Search: Map I Basic I Advanced
Browse by City: Gamnesville I Key West I Tampa

Building ID g1892_258
Contact Us I Technical Aspects I Statitsic I Privacy Policy
Powered by University of Florida Digital Collectionstechnology |F I I __ _
@ 2004- 2005 University of Florida George A. Smathers Libraries.
All rights reserved. UF IFLOR' IDA I,'F.RIT (F
Acceptable Use. Copyright, and Disclaimer Statement I'Fman R tI;EAr u 'r S T f ,.\
Last updated July, 2007 mrs
Figure 5h. Query Results

Queries, themselves, are generally limited. Query launch from the GIS
interface passes a unique building identifier (see Fig. 5h) against the UF
Digital Collections/Ephemera/ Cities collections. The strength or, rather,
the weakness of this method resides in the strengths of, first, the authority
record and, second, of periodic updates to the metadata for individual
digital objects bearing related information. Authority records had to be rich
and easily enriched, preferably by automated means. Metadata updates
placed authority record numbers into searchable data, but the process
required disambiguation of subjects to ensure use of the appropriate

16 The bulk of searchable text for the Ephemeral Cities partnered institutions resides in the
UF Digital Collections. The searchable text issues, to date, has not left much hidden; but
this is changing as the partners up-grade services to provide and support searchable
The bulk of searchable text outside the UF Digital Collections is produced and held in the
digital library of the Alachua County Clerk of Courts. All of this text resides in SQL tables.
This data was collected whole rather than harvested.


final project report page 30

record numbers. Construction of authority records is discussed in
Appendix B.

Authority record construction in an age of information extraction is
something of the chicken-and-egg problem. Records are constructed
based on the authority of printed information, identified in and extracted
from resources digitized. But, for as good as information extraction
system may become, they remain rather dumb or, more exactly, literal.
They'll remain so until both authority records have been greatly enriched
and until semantic systems have matured in equal if not by an exceeding
measure.17 Differentiation of names, whether those of place, of people, of
things or events, requires significant labor. Ephemeral Cities planners and
programmers believed that initial concentration on name rich resources
such as directories, membership rolls and the like, would aid quick
construction of sufficiently rich authority records. Disambiguation,
however, proved a much more difficult chore than was believed.

Ephemeral Cities query systems faced two challenges. The first challenge
was volume. The second challenge, again, is disambiguation. History is
frequently about the big names often, the "dead-white-guys". Most
recorded history and, therefore, most of the content digitized bears this
limitation. Digitizing 2,500 resources per city is a drop in the bucket
relative to the information needed to disambiguate Name-A from Name-B.
Even in newspapers, rich in names and events, significant mentions of
other folk is sparse and, surprisingly, light on the details that aid in easy
disambiguation of subjects.18 "Taking the search out of research" one of
Ephemeral Cities' aspirations as an aid to scholars may be noble; but, it
is easily ridiculed in retrospect as naive. Ephemeral Cities planners still
see economic advantage to be gained through the construction of this
facet of the semantic web19. Indeed, this portion of the semantic web's

17 Ephemeral Cities was so named because of the ephemeral, the impermanent nature of
cities and Florida's rapidly growing cities in particular. But, the name originally came in a
flash of green. One of the planners describe the authority systems on which the project
would be built as its Oz, its Emerald City, which other planners heard as Ephemeral City.
The name change stuck. The authority system remains at its heart and its weakness,
more so than the GIS.
18 The fact that fixed and numbered postal addresses in the United States of America is a
relatively contemporary development does not help in disambiguation, for example, of
John Smith who lived on University Avenue who taught history at the University and a
contemporary, no-relation, John Smith who taught chemistry and also lived on University
Avenue, and who was jailed for public drunkenness. With few houses on University
Avenue, the postman knew everyone; indeed, almost everyone knew everyone else.
19 Semantic Web: definition see http://en.wikipedia.ora/wiki/Semantic Web

^B ^J-jI JlfL L iV.1, J L A i J L B- ii? tliikSr uLI&

final project report

construction is frequently overlooked. But, national strategies for dealing
with these issues need to be more clearly defined to realize the

.-.E emmeral Cities Project ... ....

t -. ... ... .. .. .. Mi A *-- 4, "k

Search: Map I Basic I Advanced
Browse by City: Gainesville I Key West I Tampa

Building ID g1892_258

Search Results

New Search I Refine Your Search

Your search of All Collection Groups for'Hippodrome' anywhere resulted in 117 items in
67 titles.
Table vie, BriefView Thumbnail View Sort By:Rank

1 50 of 67 matching titles

No. Title
1 Corner view of building and parking lot. Presently The
Hippodrome State Theatre
2 Four people ascending steps of old Post Office.
3 (10 volumes )
Hippodrome Theater
4 Stevenson and another man in front of old Post Office.
5 Gainesville municipal building. Presently the Hippodrome
State Theatre .
6 Gainesville, Florida: A Circle of Diversity
7 WomaNews : Gainesvlle's Feminist Newspaper. (4 issues )
8 Book of Florida

1 8 of 67 matching titles

Next > Last >1


Kesse, Erich J, 1959-

Next > Last >

Contact Us I Technical Aspects I Statistics | Privacy Policy
Poiered by University of Florida Digital Collections technology
@2004- 2005 University of Florida George A Smathers Libraries '
All rights reserved U FLO R. 4I C fl- "it I i
Acceptable Use, Copyright, and Disclaimer Statement pr-lua ns r r hr.>, .'a r1, A^ "'n i F 1 .1i
Last updated July, 2007 mys
Figure 5i. Query Results (corrected behavior)

Whatever the challenge, both for lack of available resources or the inability
to disambiguate them, the GIS/UFDC query often comes up empty
handed. One of the project's planners has observed that it might have
been better had the project concentrated on persons, things and events
associated with a handful of major building in the downtown cores of the

page 31



final project report

target cities. This would have had the effect of more effective construction
at the cost of focusing on the "dead-white-guys" of history, or, as one of
Gainesville's radical feminists observed, we "might succeed at taking the
search out of researched, but would have drawn a point on the his in
history".20 When these issues are corrected, query results find information
in the digital collections associated with the building ID sought (Fig. 5i), in
this case a building now known as the Hippodrome and formerly known as
the (downtown) Post Office, in Gainesville. And, selection from among
these results displays the various research resources (Fig 5j).

Library Catalog

Search This Colection
S-~rh All dcin
All Vod...
Full Citation
VRelatod 1-g-e
ftg Image
Zoomable Image
P-m. F-dius.1,
Uamg Th, s &te
Coht-t Us


Title: Hippodrome Theater
Related Images

~ i '

Figure 5j. Selection display (related images view)

20 Coincidentally? Maybe not. But, after this comment was made, Leila Adams, a
University of Florida Ronald E. McNair (student) Scholar joined the University of Florida
team to bring to light the history of "Radical Women in Gainesville" (RWG) at
http://www.uflib.ufl.edu/UFDC/?s=RWG. This digital collection documents a time in
history outside that covered by Ephemeral Cities, nonetheless a pivotal location at a
determining point in American women's history. RWG is now joined with "Women in
Development" (WID) at http://www.uflib.ufl.edu/ufdc/?c=WID, a growing collection on the
international role of women in social and cultural development. WID, in many ways is the
fruit of Gainesville's role women's history. Both collections now come together as
"Women & social movements" at http://www.uflib.ufl.edu/ufdc/?q=wsma.

page 32

Imagte 2


final project report page 33

A MANY FACETED GIS. While the authority system remains week, nowhere
do we deviate more from the original project plan than in the performance
of the GIS. The critical questions are these:
Why weren't more of the year sets used? All have been geo-
Why is the GIS interface so fragmented? There are 194 separate
albeit similar GIS map interfaces, one for each distinct city section
as published by the Sanborn Fire Insurance Company.
The answer to the first question is simple: time. And, it is indicated by the
answer to the second: design complexity. The project was originally
focused between the dates 1880-1910. Project programming delays gave
the geo-rectification vendor time to complete rectification of the entire
Sanborn public-domain maps. But, the design complexity did not leave
us with enough time to index and subsequently geo-referencing every
building on every map. We had only enough time to process those we
had originally committed to.

The project's GIS staff indicated that construction of the GIS interface was
labor intensive, belabored for technical reasons resulting in a complex
network of interfaces rather than a single interface. Indeed, there is a
separate interface for each contiguous city section represented on the
printed map sheets, at total of 194 interfaces for the sum of the three
years for each of three cities. This approach, again, is not that envisioned
by the project planners. It calls into question the viability of extending the
Ephemeral Cities model either while still working under ArclMS or without
rethinking the implementation of it that we now deploy.

The rationale for separation of map sheets into city sections, indeed, the
rational for keeping even map sheets separate is that the Sanborn maps,
while very exact for maps of their day, were not exact enough from sheet
to sheet to allow their merger. The map sheets and sections might have
been "stretched" to allow merger this is relatively common practice but
it would have been very time consuming, exceeding the GIS cost-share
budget of the project, even with our addition of time from a second GIS
librarian. Because a Sanborn map sheet may depict multiple city
sections, it was then argued, city sections too had to be separated from
their sheets. The resulting GIS interface is not that envisioned and is
further separated from the seamless interface desired.


final project report page 34

In any case, because the area per map sheet depicted varied from one
publication year to the next, it was also argued that map sheets could not
be used as layers in the GIS interface, one overlaying the other, as had
been planned, to illustrate change over time. It may well have been the
case that separated into city sections of differing area through time, the
sectional maps could not be overlaid (easily enough). But, our data points
(points representing buildings, for example) remain consistently placed,
fixed on known Earth coordinates, from one geo-rectified map to the next.
As seen in the Google Earth example following (see Appendix C), the GIS
might have used a contemporary map as a common starting reference.
And, if it could not have merged the map sheets and sections, it should
have layered the sheets and sections, using points of reference, over the
common frame, turning these layers on and off as needed to "walk
through time".

If we'd taken the same faulted approach to Aerial Photography : Florida
(FLAP (http://www.uflib.ufl.edu/digital/collections/flap/), our previous GIS
project, we'd have more than 88,000 individuated interfaces. FLAP, as
Ephemeral Cities, uses non-merged component "maps". The real
difference between the projects is that Ephemeral Cities includes
additional layers: building locations. This difference does not suggest why
we have so many interfaces; the belief that the user may fault failure to
merge the layers for a given city/year pairing, may not understand the
difficulty in doing so, does suggest why we have so many interfaces rather
than as many layers.

The Ephemeral Cities planners have rethought our GIS interface and
found the newly envisioned instance little different from that originally
planned. The Ephemeral Cities interface needs to be reimplemented with
the unifying logic of the common and whole city if not whole state map.
The reasons for the non-merger of city/year set layers need to be clearly
explained. And, those layers need to be nested, as they can be in Google
Earth, allowing them to be turned on and off both, in one instance, as year
sets and, in another instance, as sectional layers over time. The resulting
product should allow all historic map layers to be turned off, as well, so as
to allow data-points to be viewed against the back-drop of the
contemporary map.


final project report page 35

Build a Geographic Information Systems (GIS) Interface.

Develop a scalable project architecture employing geographic
information system (GIS) functionality to link city maps, city directories,
documents, databases, and images ofperiod cultural objects. The
system architecture will permit the integration of additional cities and
digital objects as interest in the project develops beyond the original

The Ephemeral Cities GIS interface was built on the ESRI ArclMS
(http://www.esri.com/software/arcqis/arcims/index.html). ArclMS was
chosen as a matter of convenience and market-share. At the time that
Ephemeral Cities was planned and executed there was no viable and
certainly no robust open-source or more widely available GIS solution.
Indeed, ArclMS is used by a majority of the country's university-based GIS
programs, which like those of the University of Florida maintain generous
use licenses and "seats" for relatively large numbers of users.21

The University of Florida had previously used ArclMS successfully in it
Aerial Photography : Florida (FLAP) project (http://www.uflib.ufl.edu/digital/
collections/flap/). FLAP, which provides indexed access to more than
88,000 Florida aerial photographs from 1930 through 1975, was
enthusiastically received. The collection, built in part with funding from
IMLS/LSTA, continues to draw thousands of hits each month. At the time
of project planning, FLAP, however, was relatively young. It was known
both that its popularity was causing it to suffer load problems and that its
complexity was beginning to show some wear.

Ephemeral Cities planners believed that load problems could be fixed
through computer resource enhancements and better memory
management. This proved to be the case, though the human load in
doing so limits promulgation of this solution. It requires that a dedicated
GIS unit with good Systems support be in place at the deploying

The public GIS interfaces were beginning to show wear, taxing the abilities
and intuition of the common library patron. While advanced users of GIS

21 ArclMS is also used by GIS librarian and other GIS professionals under campus licenses
at the two other Ephemeral Cities primary partner institutions: Florida International
University and the University of South Florida.


final project report page 36

understood the interface through past experience, the common user found
the interface complex, navigation difficult to intuit, etc. FLAP and
Ephemeral Cities planners believed that the interface could be simplified.
But, simplification came at a price that ultimately overtaxed those building
the interface while also trying to serve the experienced GIS user, who
presented more immediate use of GIS. The Ephemeral Cities
programmers, both on the GIS side and the Systems side, attempted to
plug the gap through additional interface programming outside the GIS.

Simplification eventually proved too complex for the limited resources of
the Project.22 In retrospect, we note that the solutions being put forward
by the Systems side a database supported, XML encapsulated, and
scripted solution is increasingly found in web services that today attempt
to enable the functions envisioned by Ephemeral Cities. Asynchronous
JavaScript and XML (AJAX) designs now underpin Google Earth, Google
Maps, and similar services. Today, the GIS side is seeking out an
alternative to ArclMS. And, the Systems side is considering abandoning
traditional GIS systems entirely.

Digitize relevant historic resources.

Select 2,500 historical objects each for Gainesville, Tampa, and Key
West. Digitize, and create standards-compliant images and metadata.

For the majority of the project's partners and partnerships, this was the
meat of Ephemeral Cities production. This facet of the project resembled
a common digitization project. Inasmuch as possible, the project asked
partners to ensure that their metadata included place and proper nouns
and standard historical museum object descriptors. The requirement
looked no more special than the particulars of a localized metadata
encoding scheme. Ephemeral Cities purposefully did not require its
partners to supply longitude or latitude for any resource as a means of

22 The University of Florida exceeded resources granted and budgeted for interface
programming. Additional human resources were added on both the GIS and Systems
sides. The GIS side made an unplanned contribution of labor in the form of a second
GIS librarian. On the Systems side, the project programmer continued full-time beyond
the original term granted. Working through extensions and, now, beyond the project
term, the programmer has given more time to this project from University funds than from
IMLS funds on a ratio of roughly 2:1 rather than the planned 1:1 cost-share split.


final project report page 37

demonstrating that commonly created digital objects could be used in the
map interface.

The project postulated that authority records could be created and
referenced by the central system to automatically enhance metadata once
received from project partners. The architecture for this part of the project
was based largely on the GATE (General Architecture for Text
Engineering http://gate.ac.uk/) from the University of Sheffield (Sheffield,
England) Natural Language Processing Group, and specifically GATE's
ANNIE information extraction component (http://gate.ac.uk/ie/annie.html).
These were chosen both in recognition of University of Sheffield's
development strength and testing and because of specific uses by
Greenstone, the data-store for the UF Digital Collections, and by the
Perseus digital library (http:// www.perseus.tufts.edu/), which first
suggested the possibilities for an Ephemeral Cities project.

Appropriate use of GATE and ANNIE, indeed, the prospects for making
any information extraction process useful is the creation of robust authority
files. More discussion of our authority files and their structure follows. In
brief, it should be said here, while the majority of the project partners saw
the digitization of historical objects as fulfilling digit object content
requirements, the programming team regarded these objects as name-rich
data stores from which authority files could be built.

Digitization targets were reached. Some of the partners reached the
targets and continued to feed objects into the core collection. The
University of South Florida, for example, continues to feed newspaper
content into the collection. The Monroe County Public Library remains
particularly proactive digitizing local Key West history, while the Matheson
Historical Center has been successful drawing supplemental funding to
digitization of its Gainesville oral histories collection. And, the University
of Florida continues to make active field use of its digital single lens reflex
(dSLR) cameras to image local architecture for project use.

Inasmuch as possible, targets included name rich resources that, it was
conceived, would allow Ephemeral Cities to become smarter through
automation. Historical objects included historic maps; city directories,
telephone directories, blue books and other membership rosters; city and
county documents; newspapers; monographs, photographs; and archival
documents and artifacts.


final project report page 38

HISTORIC MAPS. The Ephemeral Cities model can be made to work with or
without historic maps. But, historic maps enrich the experience and
enable the "walk-through-time". These maps inform the project's historic
layers: mapped objects can be seen in contemporary time and at times
contemporary to their creation or utility. In Florida, the Sanborn Fire
Insurance Company maps that have passed into the public domain offered
excellent detail and very accurate representations relative to known Earth
coordinates. They afforded building footprints and information about
building materials and uses. Given a few reference points common to
both the historic and contemporary maps, the Sanborn maps were easily
geo-rectified by the Project's vendor, Geoassessment Services, Inc.
(Gainesville, Florida). Because the Sanborn maps were used to
assesses and insure the value of property, they were redrawn periodically.
They afforded the project with snapshots of city foot-prints over time and
in tremendous detail.

Over the course of the project, working through extensions and
programming effort to secure a working GIS interface, the University of
Florida, using its own funds,23 geo-rectified all of the public-domain Florida
Sanborn maps: all years of all of the Florida cities, in anticipation of
extending the Ephemeral Cities model throughout Florida. In 2007, the
University of Florida became responsible for maintaining the properties of
Historic Saint Augustine, Florida. The Sanborn maps of St. Augustine
are now being used to reference Historic Saint Augustine's historic
preservation files. These files provide a dense record of architectural and
social structures in America's "first city" and, it is hoped, will be used to
draw references to other colonial cities in the Americas.24

MEMBERSHIP ROSTERS. Reconstructing history required both a sense of
place and actors. Ephemeral Cities argued that while history's actors may
indeed be those most frequently studied, their lives and the worlds they
occupied involved extras. To understand historic individuals and events,
one had to have a broad understanding of causal forces and secondary

23 The University of Florida expended more than $40,000 of its own funds to complete geo-
rectification of its collection of 3,167 Florida Sanborn map sheets in the public-domain.
24 In anticipation of the model's extension to explore colonial cities, the University of Florida,
with the aid of the University of the Virgin Islands (also an IMLS National Leadership
Award project recipient) and the University of the West Indies, also imaged and
attempted to geo-rectify highly detailed colonial era maps of Saint Augustine and
Pensacola, Florida; Port au Prince, Haiti; San Juan, Puerto Rico; Christiansted and
Frederiksted, U.S. Virgin Islands; and Port-of-Spain, Trinidad and Tobago.


final project report page 39

figures. Ephemeral Cities drew its characters from city directories and
other name rich resources that placed city populations. The importance of
a directory is that it names individuals, verifies their existence at a given
time, and places them at specific locations: homes, places of work and
worship, the locations of births, weddings, funerals and all manner of civic
events between.

While we continue to trust that this model of placing history's actors on the
scene is still valid, Ephemeral Cities was naive in its belief that we could
control so many actors. It became clear that the geographic of the project
for any of the target cities let alone all of them was too large. Just in terms
of shear numbers, one directory for one year in early Gainesville history
generated more than 3,000 names in the city center alone. Directories do
little, as well, to disambiguate individuals. Abraham Smith and Abraham
Smith appear to have lived concurrently yet lived fully different lives. The
model's challenge was keeping them separate while establishing their
relationship if any. As we'll see other name-rich resources aid in this task.

The greater challenge arose in traditional library practice. Names require
name authority. And, the rules for establishing name authority records
require something more substantial than mention in a directory. Even
setting aside rules, the volume of names and the labor required to
establish them was not fully anticipated during planning. A new more
dynamic means of building authority records would be required.
Ephemeral Cities set out to define such means (see discussion following).

DOCUMENTS. If reconstructing history required both a sense of place and
actors, it also required "the sense of movement".25 To paraphrase North
Carolina's emeritus writer, Doris Betts,26 its one thing to set a scene, it's
another entirely to move a character through it. City and county
documents, newspapers, monographic accounts, and archival documents

25 The Sense of Movement (London, England : Faber and Faber, 1957), incidentally, is the
title of a volume of poetry by Thom Gunn. The mention here is tangential and might not
merit mention if the volume's lead title, On the Move, did not seem to summarize the
project's current state and desired outcome:
"Men manufacture both machine and soul,
And use what they imperfectly control
To dare a future from the taken routes."
26 Speaking at a Women Writers Conference in Danville, Kentucky at Center College, circa


final project report page 40

such as personal letters and diaries give us the sense of movement. In
them we see history's actors move through historic places. What's more,
they reveal relationships that both help us build the authority files that will
support data mining and, therein, to facilitate understandings of historic

They show us, as in this fictitious example, John A. Doe bumping into
Jane B. Smith as they make their ways, daily, to places of work, leading
eventually to their marriage. Though the event was nothing special, it
connects Doe's uncle, Jeb, a farmer, to Smith's father, a furniture maker.
Jeb describes a weed-like tree and its bomb-like fruit in idle chat. Turns
out, Jeb had so many of these bomb fruits, he'd recently decided to be rid
of them by flame. Jeb presumed that their oils would extend the life of his
evening fire. But, heck if the dab-nub things actually dampened the
flames and the oils hardened in his fireplace. Mr. Smith realized later that
night that these hardening oils might replace the flammable finishing rubs
that he used in his work. And, so, North Florida's Tung-oil industry was

These resources find the story in history. Less poetically and more to the
point of constructing the Ephemeral Cities map interface, the plotting of
people geographically in association with events could illustrate research.
The could, for example, show the proximity of saloons to churches at the
start of Florida's temperance movement, suggesting where the researcher
might look for stories of drunken sailors approaching Key West's
daughters on their way out of church.

PHOTOGRAPHS and ARTIFACTS. And, we might also include recorded audio
and video here as well, though their inclusion was not originally planned
for Ephemeral Cities.27 ... Seeing is believing. So much of making
history relevant rests in making it personal, not just in finding and telling
the story but in sensing it as well. Photographs, like storey boards for
movie scripts and plays, allow us to visualize place, actors and to some
extent event. The event horizon of a play is when the lights are turned
up and actors pick up their props to make things happen. Artifacts

27 Digital audio (and their transcripts), for the period on which Ephemeral Cities
concentrated, was recently made accessible to the project through a separate grant by
the Northeast Florida Library Information Network (NEFLIN) to the Matheson Historical
Center (Gainesville, Florida). Matheson Museum, Inc. Oral History Collections
(http://web.uflib.ufl.edu/ufdc/?c=mmoh) exposes selected audio clips and transcripts to
information extraction and mapping.


final project report page 41

elaborate the scene. They help us to understand, for example, what is to
grind corn or coffee with a mortar and pestle, to appreciate the amount of
time that went into food preparation. Perhaps with the aid of additional
information, they teach us about the popularization of leisure time and the
creation of childhood as we know them today a race of Model-T
automobile against horse and buggy illustrates time gain.

Ephemeral Cities, with the aid of digital single lens reflex (dSLR) cameras,
afforded the partnership ability to move beyond our archives and libraries.
Purchased to image artifacts during community-days events, the cameras
were also turned on our cities to capture artifacts in architecture. Florida
is rapidly developing and the old falling to new; building as points of
historical reference are recognized as ephemeral. Community-days were
paired with city-portrait days. New visual documentation of Gainesville
and Key West is particularly rich; and, most of these images have been
enhanced with geographic metadata.28 We've also continued to use them
in anticipation of project extension to other cities.29

Now, digitization of 2,500 items per city, or 7,500 items in total, is a
healthy digitization project. But, as Ephemeral Cities project managers
were to discover, that number is a drop in a sea of relational events. Even
with name-rich resources, the project experienced difficulty drawing
connections. It has been suggested that it would have been more
advisable, though perhaps less initially fundable, to have restricted the
map interface to a downtown core or to specific locations. In this scenario,
however, we might find Doe and Smith marrying at the courthouse, but the
causality of marriage both in the before and the after might be lost,
languishing in resources beyond the block or building.

28 Because images lack text that can be subjected to information extraction techniques,
geographic metadata, both textual (i.e., place names and place name hierarchy) and
numeric (decimal degrees expressions of longitude and latitude) was added to image
packages. Textual metadata was built using FGDC records. Numeric metadata was
extracted both from geo-rectified Sanborn maps and from Google Earth KML place files.
29 Selective city portraits have been created both for Spanish colonial Saint Augustine,
Florida and for Spanish colonial San Juan, Puerto Rico. No IMLS funds were used for
these portraits. Walking tours and learning modules to compare the two cities are
anticipated. Multiple uses are expected. Saint Augustine products, for example, will be
used in an extended Ephemeral Cities and in the NEH/Florida Humanities Council funded
Spanish Colonial Saint Augustine: A Resource for Teachers (http://www.uflib.ufl.edu/ufdc/
?s=teachers) collection. And, San Juan products, in an extended Ephemeral Cities and
in the U.S. Department of Education TICFIA funded Digital Library of the Caribbean
(http://www.dloc.com/), based in UF Digital Collections technologies.


final project report page 42

It seems that just as digital librarians have begun to round the corner,
saying as it where, it's no longer how much you've got but what you can
do with it, that we find the truer adage rests in scan more and do more
with more meaningful stuff drawn from content wherever it may reside
(i.e., in more than one collection, at more than one institution). It becomes
advisable to build more collaboratively, to ensure that networks of content
contributors offer researchers the most robust resource. (More
discussion of our collaboration follows.)

The digital objects created or enhanced as a result of Ephemeral Cities
were deposited into a number of digital collections, most under the
umbrella offered by UF Digital Collections (UFDC) but also under that of
PALMM or in the more localized digital libraries of the partners. Several
resources were duplicated under each umbrella. Content deposited with
UFDC was further distributed, placed into topical collections to facilitate
self-directed research of particular patrons. The Sanborn maps, for
example, fell into a named collection, the Sanborn Fire Insurance Maps
of Florida (http://www.uflib.ufl.edu/ufdc/?c=Sanborn). Photographs and
artifacts found their way into one of the Florida Photographs collections
(http://www.uflib.ufl.edu/ufdc/?q=foto) and most frequently into the Florida
Ephemera collection. Other placements facilitated access to Florida law,
oral histories, Florida maps, Florida newspapers, etc. Content deposited
with PALMM was further distributed into, principally, the Florida Heritage
Collection (http://palmm.fcla.edu/fh/), but also into Floridiana on the Web
(http://www.lib.usf.edu/ldsu/digitalcollections/F03/html/), and Mile Markers

The Ephemeral Cities application was intended to use distributed queried
methods to discover content on allied digital libraries. The method,
particularly when paired with the UFDC's (Ephemeral Cities') Greenstone
core, was insufficiently robust. The method, when partnered with
metadata from the DLXS data store (PALMM's core), proved somewhat
incompatible. Inasmuch as possible, Ephemeral Cities has ingested either
metadata or whole packages (i.e., metadata and digital objects, essentially
mirroring localized partner libraries) to better ensure discovery. Discovery
on the whole, either because of query limitations or large but still
inadequate content, is disappointing at the level of a building's footprint.
Ephemeral Cities planners are hopeful that continued content construction
and future developments of query methods will improve discovery.


final project report page 43

Invite local, general participation in and contribution to Ephemeral

Hold a "My Town" event in each of the three cities. During this event,
citizens will be encouraged to bring in their own historic period
artifacts for digital capture and sharing as part of this project.

EVENTS. My Town events, also known as Community Days, were
envisioned as community building events. Usually carried out with the
assistance of the local public library's publicity arm, these events invited
citizens to become part of the project. On their face, these events asked
citizens to, as Monty Python might have called out, bring out your dead,
bring forth artifacts illustrative of life during the Ephemeral Cities target
decades for imaging, appraisal and preservation advice. Ephemeral Cities
planners expected to hear the rejoinder heard in the Monty Python skit:
but, I'm not dead.

To ensure that we captured context solicited from those who participated,
we asked participants to tell us the stories of their artifacts. Everyone who
participated had stories to tell and very much enjoyed telling them. The
planners opted to record this information in writing rather than by
recording for fear that recording would complicate the event. While the
stories told were far more rambling than those gathered in the making of
an oral history, in retrospect, some of the partnerships might have planned
oral history booths similar to the National Public Radio's Story Corps
(http://www.storycorps.net/). The experience also suggested a need to
expose oral histories, previously collected but not yet digitized, for project

My Town events were somewhat disappointing in their poor attendance.
Held Saturdays with generous advance publicity in relatively good, well
known locations, the events attracted a largely aged population. Those
attending had difficulty doing so. A different sense of movement slowed
attendees and complicated their arrival with artifacts in hand. Several
made alternative arrangements to have photographers come to them.
This was particularly surprising for the Gainesville partnership. Annual
bottle days and similar collectors events at the Matheson Historic Center
draw hundreds of interested people. And, a similar albeit much broader
collectors' fair at the University of Florida's O'Connell Basketball Arena


final project report page 44

hosts hundreds of booths and attracts thousands of interested parties for
Alachua County and surrounding counties.30

In Key West, the Convent Hospital, also known as the Convent of Mary
Immaculate, scheduled an imaging session dedicated to its own holdings.
This experience led to what became known as the City Portrait, which
gave the photographers additional days and allowed them to travel across
Gainesville and Key West to image architecture, monuments and other in-
situ artifacts. Some of the Ephemeral Cities partners approached the City
Portrait with peculiar zeal and volunteers for those partners continue to
contribute images to map-out the collection with architectural artifacts,
many representing building surveys of considerable detail and frequently
accompanied by longitude and latitude and, less frequently, with brief
historical narratives.

each of our partnered groups, Ephemeral Cities developed collaborations
that brought materials and talents together, encouraged by the Institute for
Museum and Library Services' (IMLS) requirements for this National
Leadership Award. Each of the partnerships offered a different test-bed.

When Florida's digital collaborative, PALMM, was being planned a decade
ago, planners suggested that Florida should grow building strong regional
leads that would reach out into their communities to build capacity and
feed content into the collaborative. This objective was met with varying
degrees of success. It may be fair to say that this PALMM objective was
met only in areas where development was driven by the engine of IMLS
funding and its requirements for collaboration. Ephemeral Cities'
planners, recognizing the need for more broadly representative if not more
comprehensive digital content, encourage IMLS to emphasize similar
collaboration in its State Programs (http://www.imls.gov/programs/

One of the Ephemeral Cities (local) partnerships provided a strong central
service that almost wholly met the needs of its partners. This was not
what the Ephemeral Cities' planners had intended. While strong

30 Objects displayed at the University of Florida event were, for the most part, out of scope,
from beyond Ephemeral Cities target years. Each of the partnerships might have been
more aggressive in making arrangements with local antique dealers, but we sought to
avoid issues of promoting businesses on our not-for-profit sites. (At the University of
Florida, where Ephemeral Cities is housed, web policy prohibits such relationships.)


final project report page 45

centralization ensured completion of work, it did not grow capacity in the
community served. When the project funding had been consumed,
community contribution ceased. The central service, however, did
continue to grow and is now launching production-level services that may
yet again reach beyond that partnership's lead institution.

This model, driven largely on available funding, suggests consideration of
fee-for-service structures. Those needing or making use of the collected
digital objects, academics and K-12 teachers as well as the life-long-
learners in our communities, are unaccustomed to paying (and generally
not sufficiently funded to afford to pay) for the product they receive. Even
if they were able to pay, it is doubtful that they would pay in advance or
commission photography and digitization to meet their needs. Indeed,
most of these constituents work within academic and public library
environments which traditionally cover their needs through purchase of
commercially available digital collections. The model certainly pays the
bills; it makes digitization sustainable but may not (quickly) meet the
research needs that drive academia.

The two remaining partnerships distributed labor relative to the talents and
strengths of the individual partners. In each, one set of partners, with
strong collections and an increasing pool of imaging talents, and another,
with strong metadata talents and direct links to a large digital object store
and digital archives, found commonality in content-for-technology
exchanges. Theirs was the model of collaboration that Ephemeral Cities
planners had anticipated. But each was sufficiently distinct as to consider
them separate models.

One of the two exercised a strong sense of collaboration. Principals in the
dyadic relationship coordinated partnered activity. The content
contribution core, here, was especially strong reaching out to nearby
cultural institutions, sharing access and building capacity with a
commonality of purpose. The fact that the strong partner on this side of
the dyad was a public library rather than a university may have had
something to do with the partnership's success. None of the universities
among our partnerships was a lead-force in their community for
community content gathering. The university in this particular partnership
was a strong metadata proponent, however, and ensured that resources
were appropriately readied for research uses.


final project report page 46

Relative to economic forces, the community content component of this
local partnership seems to have outlasted its Ephemeral Cities funding. It
could be said that the sense of community and the sense of service to the
community has sustained it. But, without infusion of continuing funds, it
moves forward slowly. The difficulty with this model is that a slow-down
for one party, in many ways, slows the other party as well. While neither
party was slowed for the duration of Ephemeral Cities funding, both
principal parties now move forward slowly. In this dyad however, it is the
university partner that comes out the weaker.

Ephemeral Cities and PALMM planners envisioned that university library
digitization programs, driven to meet digital access demands among
researchers, would always remain relatively strong, i.e., stronger than
digitization programs among cultural institutions in their local region. And,
many of us believed that technology sharing as a means of building
content for university researchers would be a strong motivator for
collaboration and continued funding. Here is an exception to what,
otherwise, has been proven a rule. Nothing in the Ephemeral Cities
experience suggests a cause.

The second of the two exercised a strong sense of institutional
independence. The institutions of this partnership had a long history of
independent granting and digitization projects. Each, by the time of
Ephemeral Cities launch, offered services considered to be integral to
their institution's mission. Many had built strong volunteer programs to
afford project prototyping in advance of granting or to tide them through.
Some had built endowment that served, at times, as a buffer for
continuation beyond project funding. But, while the partners continue to
collaborate with one another from time to time, perhaps because of its
spirit of independence, the partnership did not develop the sense of
community. In insulating itself from economic factors it works without
continuing common goals; but, this is not to say that it lacks common
cause. The weakness of the Ephemeral Cities application in distributed
query or in harvesting/gathering digital content is its greatest weakness.

This is a weakness not solely limited to Ephemeral Cities. Our digital
library harvesters need to take in our text not just our records with their
pointers out to distributed and individually searched textual digital
libraries.31 Our records need to tag geographic and temporal information.

31 While Priscilla Caplan's recent editorial, "Ten years after" (Library Hi Tech. v.25, n.4
(2007), pp.449-453) focus on issues in the field of digital preservation, the goals and


final project report page 47

And, both our records and data-mining/information extraction methods
need to be fully supported by ontologically structured authority files. The
researcher is still forced to do too much searching (in too many locations).
And, these data cores even Google's need to become more capable of
facilitating complex searches. The prospects of the semantic web still
require considerable construction. Research methods need to be better
studied and more accurately emulated. The economic gains, in research
time alone, stand to be significant.

SUSTAINABILITY OF CONTENT. In each of Ephemeral Cities three
partnerships, the lead institutions is a state university Florida
International University, the University of Florida, and the University of
South Florida with access to the Florida Center for Library Automation's
Florida Digital Archive (FDA).32 FDA is supported by the State of Florida
with continuation funding in the budget of the Florida Center for Library
Automation. It was originally designed and programmed with funding, in
part from the Institute for Library and Museum Studies. And, it is widely
recognized both in the United States and worldwide, as an Open Archival
Information System (OAIS)33 based PREMIS-compliant34 model digital

Because FDA is, functionally, relatively new and still working through a
large backlog of accumulated content, the lead institutions are still in
process of committing their Ephemeral Cities content to FDA. Each of the
three institutions is committed to digital archiving. Content not yet
committed to FDA is generally backed up, until it can be properly archived,
on burn-verified DVDs or on a Storage Area Network (SAN) with
redundant-backup, logged and tracked in the digital asset management
systems (DAMS) of the lead institutions. Because a large portion of
content is duplicated on the project's central systems at the University of
Florida, additional redundancy exists for that portion duplicated.

strategies to which she points should be focused as well on enhancing the utility of digital
collections. If a researcher could focus more on the analysis of findings than on the
search for them, our academic (and economic) capital would sore.
32 Florida Digital Archive (FDA), see: http://www.fcla.edu/digitalArchive/
33 Open Archival Information System (OAIS), see: http://www.oclc.org/research/
publications/arch ive/2000/lavoie/
34 PREMIS (PREservation Metadata Implementation Strategies), see: http://www.loc.gov/
35 FDA is not yet a fully certified digital archive. The results of a pilot audit conducted by the
Digital Curation Centre in 2007 are available at http://www.fcla.edu/digitalArchive/pdfs/


final project report page 48

FDA systems report archiving activity back to the institution for archiving
particular Ephemeral Cities content. To ensure audit of responsibilities
and that no digital objects, no packages fall between the cracks, are lost in
transmission or after delivery, the University of Florida programmed a FDA
Reader application. The application, currently server-based has passed
tests and is now being modified for distributed use by the other lead
institutions as well as other institutions using FDA services.36

Data archived with FDA is redundantly stored with the University of
Florida's Computing & Network Services (CNS) facility in Gainesville,
Florida and with the Northwest Regional Data Center (NWRDC) on the
campus of Florida State University in Tallahassee, Florida.37 All digital
objects, including digital masters and derivatives images, text, audio and
video files, as well as corresponding metadata, including bibliographic,
geographic, temporal, and structural data, are archived. And, archive
policy outlines migration strategies and procedure appropriate to the
archived objects.

SUSTAINABILITY OF APPLICATIONS. The Ephemeral Cities project hired a
project programmer originally funded 50:50 using funds from both IMLS
and the University of Florida.38 This was conditioned on continuing
employment of a programmer to sustain the project; a programmer
continues to be employed, albeit with expanded responsibilities, and both
project and supporting applications continue to be maintained.39
Additionally, the Geographic Information Systems (GIS) map interface

36 The application will become available to, among other institutions, the University of
Central Florida and its Central Florida Memory (http://www.cfmemory.orq/), another IMLS
National Leadership Award funded project. At this time, there is only the presumption
that other institutions will use the application. The application may or may not be offered
to the Florida Center for Library Automation (FCLA). We believe that audit credibility is
better insured if not offered to FCLA.
37 More information about the Archive is available at http://www.fcla.edu/digitalArchive/.
38 A Grants Office error actually moved the project programmer to University of Florida
funds early. The funding ratio was approximately 40% IMLS funds and 60% University
39 The University of Florida Libraries' Systems Department now retains two programmers
supporting the Digital Library Center, which includes, in part, the Ephemeral Cities
project. Responsibilities are divided: one programmer maintains digital object creation
support and digital asset management systems; the other maintains project interfaces
and information processing applications. There is considerable cross-training between
the senior and the junior programmer. We believe that this ensures a level of service


final project report page 49

component of the project called upon the University of Florida to cost-
share a portion of the GIS Librarian's time to build and maintain the
interface. The University of Florida continues to employ a GIS Librarian
who maintains responsibility for the Ephemeral Cities GIS map interface.40

The University of Florida foresees no change in level of support, as
originally required, into the future for either the programming or the GIS
components. Applications, hardware and services are routinely
maintained, each with redundancy.

Create education modules building research skills.

Create 10 educational modules based on appropriate Florida Sunshine
Standards [http://www.firn. edu/doe/menu/sss. htm] and the National
Geography Standards [http://www.ncge.org/publications/tutorial/
standards/]for use in classrooms and independent use by interested

Delays in completion of the GIS and other project components, pushed
the project off schedule. One casualty of delay was the window opened
for the creation of learning modules. The project plan originally called for
their creation by the University of South Florida's Florida Center for
Instructional Technology (FCIT http://fcit.usf.edu/), the State's best
regarded and most active center for instructional technology. Both project
delays is FCIT's success in granting made completion of education
modules by FCIT impossible. The work was shifted to the University of
Florida's Digital Library Center. The Center has completed the full
compliment of ten modules and continues to produce modules for specific
class-room uses. The following modules are complete.

Ephemeral Cities: Cigar Box Art (
edmod files/UF/O 1/03/01/85/00001/EDhemeralCitiesCiqars.pdf)
Ephemeral Cities: Virtual Explorations in Changing Landscapes
and Neighborhoods ( files/
UF/06/01/07/71/00001 /EphemeralCities. pdf)

40 The University of Florida Libraries' GIS Program now retains two GIS librarians, ensuring
service redundancy.


final project report page 50

History's turbulent winds and waters : historical hurricanes
( files/UF/05/09/04/72/00001 /
How to Read Sanborn Maps in Ephemeral Cities
( files/UF/01/04/01/85/
00001 /HowtoReadSanborn. pdf)
Miami: from stone circles to high rises (http://www.uflib.ufl.edu/
ufdc/UFDC.aspx?c=edmodl&m=hdlJ&i=41544) this is an Aerial
Photography : Florida education module, but makes use of the GIS
Navigating Ephemeral Cities, Part One: Alternate Routes for
Finding Information ( files/
Navigating Ephemeral Cities, Part Two: Alternate Routes for
Finding Information ( files/
Navigating Ephemeral Cities, Part Three: Alternate Routes for
Finding Information ( files/
UF/01/03/01/83/00001/Using EphemeralCities3. pdf)
Navigating Ephemeral Cities, Part Four: Alternate Routes for
Finding Information ( files/
The News in Different Places and Times (
gsdl/collect/fdnll files/UF/04/01/09/76/00001/News.pdf) this
module makes use of news articles and reporting; it was created for
Ephemeral Cities but without specific mention so it could be used
as well with the Florida Digital Newspaper Library, the home of
newspapers digitized for Ephemeral Cities.

Research 101 : Researching and using online materials to see
Tampa over time ( files/

Strolling through a college town : Gainesville and the University of
Florida ( files/UF/06/01/
07/73/00001/EPCGainesville. rdf)


final project report page 51

Ties that Bind : Key West in History (http://www.uflib.ufl.edu/ufdc/
UFDC.aspx?c=edmodl &b=UF00076714&v=00001)

Tracking through Florida : Florida's Railroads (
gsdl/collect/edmod files/UF/O 1/04/01/84/00001/Railways.pdf)

Promote the model.

Promote the use ofEphemeral Cities and facilitate the development of
similar atlases for other Florida cities and for similar efforts in other

If we did one thing well, we promoted the model as planned and described
the need for the map interface, what we called the "geo-temporaf' core for
historical studies.

Our Ephemeral Cities presentation is available as Appendix A. The
presentation was delivered as Ephemeral Cities: reviving Florida's past at
WebWise 2004; a recorded version of the presentation is available online
at http://www.uic.edu/depts/lib/systems/webwise2004/session1.ram. A
version was also published as Ephemeral Cities in: RLG DigiNews (Vol. 8,
no. 6 (2004 December), online at http://diqitalarchive.oclc.org/dal
article. Other presentations of the project, its concepts or its goals were
present at the SOLINET Annual Membership Meeting 2004 (Atlanta, GA.
2004 May); at an IMLS Virtual Cities Projects Review hosted by the
University of Kentucky (Lexington, KY. 2004 August), at the Readex
Digital Institute hosted by the Readex Corporation (Burlington, VT. 2004
October) and at the 2005 annual meeting of the Association of Caribbean
University, Research and Institutional Libraries (ACURIL) (Trois-Islets,
Martinique. 2005 June) where it was delivered in print and verbally in
English, French and Spanish (see, Appendix A.)

The project was also mentioned a presentation entitled Typeset with the
Aid of Lawn-mower the Florida Digital Newspaper Library at the 2007
annual conference of the Florida Library Association Conference (Lake
Buena Vista, FL. 2007 April), where the utility of newspapers' geo-
temporal data was described.


final project report page 52

The importance of the geo-temporal core in resource discovery became a
drum-beat of our discussions in digital library circles.

Model or Not? What we can recommend

No excuses. Ephemeral Cities never came together completely. Its
planners envisioned a service similar, in its GIS interface, to Google Earth
before Google Earth was a vision. Google Earth, which leap-frogged us
with a much larger development budget, does a better job of seamless
design and object integration. We cannot recommend Ephemeral Cities
as a model at this point in time. However, we can make the following
research and practical recommendations.


THE GIS. The underlying system of GIS support chosen by Ephemeral
Cities, ArclMS, was and remains too limited for deployment of Ephemeral
Cities as a model project elsewhere. Selection of ArclMS, recognized for
its market-share among universities, the institutions most likely to adopt an
Ephemeral Cities model, weighed system costs part of the campus
contract more heavily than its limitations. A project like Ephemeral
Cities is relational by nature, a GIS needs not only to use relational
databases but to be able to adapt to various SQL record structures or
XML data expressions.

Replace ArclMS with GIS or GIS-like (i.e., an AJAX)
technology supporting relational databases and extensible
mark-up languages. Inasmuch as possible, this should be
a national initiative. Standards-based designs for data
structures and component programming should be widely


final project report page 53

It is good to see ESRI, the makers of ArclMS, utilizing the KML format.41
Good too, to see the National Atlas Map Maker,42 also new since
conception of Ephemeral Cities, demonstrate a more useful, more robust
and more flexible ESRI-based GIS. But an Ephemeral Cities interface
must also be able to read an XML authority record, a METS record and
other record structures commonly supporting digital libraries and digital
objects. The National Atlas does not appear to read XML encapsulated
data; it retains the preference of earlier GIS for data tables. But, that isn't
to say that its Javascripts and ASP, its Internet presence could not be
coded to extract data/information from XML files for use in the GIS.

Google Earth may or may not be a panacea. It is comforting if not
encouraging to see NASA, for example, building on Google Earth
technology.43 Google Earth is more than KML, and its AJAX APIs are
openly available for development. But, Ephemeral Cities has taught us to
temper our idealism and enthusiasm. In August 2007, Google announced
"a new layer in Earth that allows you to explore locations through the lens
of the world's books."44 Imitation is the most since form of flattery. We're
flattered that an idea conceived among the Florida partners and funded by
IMLS, was also conceived of and funded by Google. Short of vindication
for our failures, however, we're not sure what Google's success might
indicate. The Google Earth "Book Search" layer was short lived, a
footnote in the blogosphere.

AUTHORITY SYSTEMS. We're just guessing, but our experience suggests
that Authority Systems, that this part of the Semantic Web, remains the
weakness of "Book Search" in Google Earth as it does in Ephemeral
Cities, or as it would were the ESRI products to mature to accommodate
information extraction and deep links into textual documents and
descriptive metadata. More attention needs to be paid to the building-out
of authority records.

Traditional MARC authority records45 are far to narrow. They relate
names and provide referencing information accommodating dates and
relationships; but, they don't make this information searchable in a

41 ESRI ArcGIS Explorer (http://www.esri.com/software/arcgis/explorer/)
42 http://nationalatlas.gov/natlas/Natlasstart.asp
43 NASA World Wind (http://worldwind.arc.nasa.qov/)
44 Google Book Search in Google Earth, see: http://qoogle-latlong.blogspot.com/2007/08/
45 See: http://www.loc.gov/marc/authoritv/


final project report page 54

meaningful way, let alone with the nuances needed to facilitate information
extraction, machine aided disambiguation, and discovery.

Evolving standards for authority within the archival records community
delimit nuances. It's Encoded Archival Context (EAC),46 an
implementation of albeit, a variation from the International Standard
Archival Authority Record (ISAAR),47 provides a fuller record. But, neither
EAC nor ISAAR is fully delimited (or sufficiently relational), as yet, for use
in information extraction. Work ongoing with North Carolina ECHO's
North Carolina Encoded Archival Context Project48 is designed to facilitate
creation of EAC records in machine-aided production environment.
However, it has not yet been awakened to, re-purposed for automated
information extraction as a production tool and not yet optimized for post-
production use in discovery.

Both MARC and EAC (as well as ISAAR and authority record constructs
within TEl and other schemes) are constructed largely for human use and
limited computerized uses. A typical authority record, however composed,
wherever it resides, is typically designed for a publication format, intended
to be read by human eyes, and to be used for disambiguation without
machine-assistance. Records are lightly tagged and most known
information is recorded in narrative strings. This is not to say that these
records cannot be used to disambiguate persons and places; they can.
But, revision of these standards, or creation of a wholly new standard for
use by code, in programming, is needed to facilitate machine-assisted
disambiguation. Such action will build more complex, even heavily tagged
records to be certain. But, these more complex records will not allow
known information to go to waste, unexposed to automated semantic

Review existing authority record standards for restructuring
as fully delimited, relational and multi-language tools. Re-
engineer authority records for use in automated information

46 See: (a) Yale University: http://www.library.yale.edu/eac/, (b) the University of Virginia:
http://www.iath.virginia.edu/eac/ and, for the most recent work, (c) North Carolina
ECHO's North Carolina Encoded Archival Context Project: http://www.ncecho.org/nceac/.
47 See: 2nd Edition as adopted (October 2003): http://www.icacds.org.uk/enq/ISAAR(CPF)
2ed.pdf or as finalized for publication (2004): http://www.ica.orq/sites/default/files/
48 See: http://www.ncecho.org/nceac/; see also, North Carolina Biographical and Historical
Information Online (NCBIO) at http://www.ils.unc.edu/digitalnc/ncbhio/index.htm


final project report page 55

extraction, semantic engineering, and machine-assisted
disambiguation of textual data.

Relative to place names, the U.S. Board on Geographic Names (US BGN
or BGN), maintains the most comprehensive authority records more
than 2.5 million names for the United States of America. Its Geographic
Names Information System (GNIS)49 provide name information,
geographic positioning data, coding references, feature characterization,
etc. GNIS provides some support for alternate and historic names, but
little delimitation of historic relations. And, while the GNIS database is
extremely large, it often lacks local and particularly ephemeral names.

Set in place initiatives to augment GNIS data at state and
local levels with ephemeral names. GNIS is already
greatly augmented by many local, regional and state
authorities. Means of collecting, vetting and approving
names should be established.50


If a full-fledged Ephemeral Cities project is now not fully possible or too
difficult, too costly to implement, local institutions and regional
partnerships can begin to prepare digital resources for use by an
Ephemeral Cities project in future. The following recommendations are
not inclusive of one another. They may be followed separately or together
in any combination. There is no particular order to these
recommendations. These recommendations could be said to describe
methods of uncovering the dots. Connecting the dots, whether in a GIS or
other application, can come later.

49 See: http://geonames.usgs.gov/domestic/index.html
50 We believe that GNIS has methods in place to do this in some measure. We believe,
however, that those methods may be inadequate to deal with the numbers of historic -
many ephemeral names to be recognized. GNIS policy (http://geonames.
usgs.gov/pppdgn.html) makes difficult the establishment of historic names, particularly
those no longer in local use. We recommend that GNIS investigate and implement
something similar to the Name Authority Co-op (NACO) at the Library of Congress.
Currently naming suggestions are sent upward to a naming authority in GNIS. We
recommend that some of this authority by shifted downward to (some of) the state name
authorities which now pass suggestions upward.


final project report page 56

RECTIFY MAPS. With varying levels of difficulty, exactness and cost, every
map of any real place can be geo-rectified. But, not every map deserves
it. Highly exact, name-rich historic maps that outline the ephemeral detail
of cities such as the Bromley, Hopkins, Hyde, Sanborn and Perris maps
in the United States of America or the Booth, Goad and Hyde maps in
Canada, the United Kingdom and parts of the British Commonwealth are
deserving. These are the types of maps that researchers will frequently
consult to fix location, conduct proximity studies, and harvest
supplemental data. These maps can be used as historic layers atop the
base map of contemporary cities.

Digitize highly exact, name-rich historic maps (e.g., fire
insurance maps, real-estate valuation maps, etc.) as 24-bit
("true color") images, at not less than 300 pixels per inch,
and, no reduction. Save the image as an uncompressed
If color is truly important, as it is for most of these maps,
image correction should be minimal and manipulations
based on firm knowledge of the as published state and
color formulations used in printing.
As possible mask the back-ground (i.e., paper) color,
saving the image as a supplemental uncompressed
layered TIFF (in addition to the original uncompressed
unlayered (TIFF).

Geo-rectify digitized maps, fixing map corners and not less
than 5 internal points on the map to known Earth

REFERENCE MAPS. Name-rich maps provide historic reference for known
locations. Extraction of names from these maps will enrich data locating
services. Even if the maps can't be used in digital space or geographic
information systems, reference information can still help the researcher fix
location and determine geographic or relational (causative) factors.

51 Fixing points is more difficult than might seem, particular for ephemeral features of
historic maps. Use trained staff to rectify maps. Knowledge of the location to be rectified
is beneficial as is access to other previously rectified maps and data.


final project report page 57

Index names appearing on (name-rich) maps.
As possible, index names using a (relational) database or
spreadsheet for the collection and arrangement of
associated data, for example: building name or reference;
map date; map sheet (page) reference; map set reference
(bibliographic identifier); building use (using standardized
thesaurus); building characteristics/materials; street name
or reference; street number or reference; etc.

Geo-reference indexed names, presuming the map on
which the names appear has been geo-rectified.
As possible, use a database or spreadsheet for the
collection and arrangement of associated data, for
example: building name or reference; point data, including
longitude and latitude; and/or polygon/footprint data,
including as series of coordinates.

Integrate indexed or geo-referenced information into
established gazetteers, or, build a local supplement to a
gazetteer of choice if the gazetteer is closed.

as well as life-events records (birth, marriage and death). Names aren't
hard to come by, but each is difficult to fix in place and time. Name-rich
records such as these help to associate a name with dates, places, and
other data such as occupation, political affiliation, race, and religion. Life-
events records begin to build relational information.

Digitize name-rich resources that fix dates, places and life-
events at not less than 300 pixels per inch, with no
reduction, minimal image manipulation, and to the bit-depth
most appropriate to the source document. Save the image
as an uncompressed TIFF.
Digitization should be optimized for highly accurate optical
character recognition (OCR) or successful transcription
(double-keying) project.


final project report page 58

As possible, OCR or transcribe these resources. Note that
an OCR/transcription project on this type of material has
little value if not accurately converted. Accuracy
confidence should be 99.95% or greater. Text should be
corrected if lower.

Parse these name-rich resources into an authority system.
As possible, the authority system should be able to
accommodate all of the information you have collected.
(Again, no authority system in current wide-spread use fully
delimits/tags information as richly as we recommend that it
be delimited.)
As possible augment your system of choice with
"supplemental" data data that the system's standard
does not accommodate using a (relational) database or
spreadsheet and referencing the authority system's
authority record number.

person, place or thing is important is knowing history and understanding
context, causes, reasons, etc. Populating a digital library with archives
and newspapers is the closest we come to time-travel. This is the story as
told by someone who was close at least temporally to it. This is where
interpretation and sometimes deceit begin, where the town being painted
red can be seen.

Most digital projects start with secondary sources. They're easy to
handle, and easy to describe. But secondary sources are like second
helpings; they're somewhat cold and a little stale. History has already
been colored; indeed it's been re-colored, tinted by an author's
perspective, misunderstandings, misinterpretations and, sometimes,

Digitize archival resources that fix dates, places and life-
events at not less than 300 pixels per inch (600 ppi,
recommended for handwritten documents), with no
reduction, minimal image manipulation, and to the bit-depth
most appropriate to the source document. Save the image
as an uncompressed TIFF.


final project report page 59

Digitization should be optimized for accurate optical
character recognition (OCR) project.
As possible, transcribe these resources and perform
quality control on the product to ensure the accuracy of
names and dates.

Digitize historic newspapers as 8-bit images (or 24-bit if the
page contains meaningful color), at not less than 300
pixels per inch (400 ppi, recommended by the U.S.
National Digital Newspaper Program52), with no reduction,
minimal image manipulation. Save the image as an
uncompressed TIFF.
Digitization should be optimized for successful transcription
(double-keying) project.
As possible, OCR these resources and, as funding allows,
perform quality control on the product to ensure the
accuracy of names and dates.

Extract names and date information from these resources,
parsing them into an authority system. (See also,
Recommendation P7 and Appendix B.)
As possible, use information extraction technologies to aid

SUPPLEMENT or ENRICH AUTHORITY. Without delving into the complexities
of GIS or other mapping systems, begin to see the dots in context. Like
the theory that underpins the social networking concept of "six-degrees-of-
separation," all data occurs in a network of connections. The following
recommendation encourages augmentation of authority records in
preparation for use by information extraction and analysis systems as well
as mapping systems. Collection of this information aids disambiguation
processes whether unaided or aided by automation.

Supplement or enrich known data.
As possible, for a given authority record enhance it data
gathered from external sources.

52 See: http://www.loc.gov/ndnp/ for conversion recommendations.


final project report page 60

An historic place name extracted from a map might be
enhanced with a current place name, coordinates, names
of associated individuals and events, transition and other
important dates, etc.
A personal name might be enhanced with middle name(s);
aliases; life event information including life event
characterization and dates; names of associated
individuals and their relationshipss; etc.


final project report page 61

Appendix A : The Ephemeral Cities Road Show
Presentation Narrative.

This is how I usually begin. Kind'a boring isn't it? Nonetheless, these are
the project's objectives:

To create an historic atlas of Florida;
To link historical resources to that atlas, providing new geo-
temporal methods of discovery;

And, finally,

To construct learning modules that would acquaint K-12 students
with these methods.

Ephemeral Cities is a technology project rather than a collection building

I would ask you just for a moment to close your eyes and envision
"Florida". What is it? What kind of picture have you formed? Some of
you, particularly those of you from southern Georgia have pictured
swamps or mangroves.


Create an historic atlas

Link historical resources
.3 bu d ;urponr;
Author learning n,.odules,


^B ^Ljj-J -L- i .,J JY^ ~ ^ LJ --L i? twi&k-r^uL&

final project report



And, for some of you, visions of gators are dancing in your heads. Still
others of you, have formed visions of mermaids! (It helps to have eaten
little and been at the bottle early for this neat trick to work.) At least, it
works well for many a college student visiting the Sunshine State on
Spring Break. But, most of you have conjured up deep and fading
memories of Florida vacations. ... Some cut short by hurricane.

page 62


final project report page 63

We Floridians are confounded daily by visions of Florida, well-surpassing
this brief survey. History hasn't been kind to us, either. Florida had been
Spanish, French, Spanish, British, and, Spanish again before becoming a
U.S. territory. Between times, we maintained multiple personalities. Even
today, we wave supplemental heritage banners including the State's
winter-highway flag.

And, with 9 of 10 of us from some place else, the question arises: "Who
are WE?" Just this past weekend, I was mistaken for a German: "What
makes you think I'm German?" I asked. "Well", responded the AIITel clerk,
looking at my name, "you've got an accent!" Determined not to say a
word, at the Latin market next door, I collected groceries, in silence. So,
what do you think the clerk asks? ",De d6nde es usted? ,Brasil? she
inquired! "Where AM I?" I thought. The trouble is, most of us don't know
what Florida is.

We, Floridians like to think ourselves as part of the United States and at
one with the American South. To ensure that our vision is shared, we
share our digital content with AmericanSouth.ORG (http://www.
americansouth.org/) in numbers as great as those of Georgia or North
Carolina. Yet, we know that may a southerner doesn't accept us as kin.
"Damn, Yankees!" We hear it all the time. Why, I even heard it just
yesterday as I drove myself to the airport.

But, while Atlanta may be the capital of the new South. Miami, if you
haven't heard, is the capital both of the Caribbean and of Latin America.
It's enough to make one's head spin! ... even without the tea-cups of


final project report page 64

When we looked at Florida through the eyes of AmericanSouth.ORG, we
saw that there was no wholly there there. Ephemeral Cities is our attempt
to anchor our communities in place, and, to demonstrate a geographic
proto-type for understanding Florida. But, we also took note of Kate
Nevin's3 observation that AmericanSouth.ORG was, yet, too much the
reservation of southern universities.

Florida's digital collaborative, PALMM (http://palmm.fcla.edu/), had
purposefully begun as an effort of the state universities. We, first, would
get our bearings, build the infrastructure of collaboration among ourselves,
then turn it loose in such a way as each university would build
partnerships within its region, ... each partnership, deepening the local
representation in PALMM Collections. Ephemeral Cities marks the time
when that time had come.

53 Kate Nevins is SOLINET's Executive Director. Her observation was voiced at an
AmericanSouth.org planning meeting (Atlanta 2003).


final project report page 65

When I moved to Florida, an elderly gentleman would grab me on my
nightly walks around the neighborhood that once had been the University
of Florida's campus. He would hold me by the elbow and guide me from
block to block as though I might have been naughty child.

In front of each house, he'd shuffle to a stop. "Miss Emma lived here
when I was a boy," he might say. "She was an upright woman. 'would
boll-up peas she, herself, had planted." He'd gather-up a breath as if
tasting history. "Don't ya know, her husband, Jimmie, was a no account
..." And, of course, I didn't know.

The marvel of those walks, night after night traveling the same blocks, was
that the stories were never the same, never repeated. Each night, he
would bring the neighborhood's characters back to life in a sequence
spread like peanut butter over time.

I hope you'll forgive me if I don't detail the importance of Miss Emma or No
Account Jimmie in the limited time I have available today. We believe that
the common characters of history may have something to tell us. Think of
their stories as street-theater that how might a playbill put it is "A

President B enters a grocery store, encounters clerk C. A few short words
later maybe about the price of milk or barcode scanners and history
has changed, forever. I assume that you all recall the story of the first
President Bush's campaign foray into a supermarket. The event is
recorded in newspapers and even in books but, wouldn't it be interesting
to compare, say, the President's letters to the First Lady on the day's
events with the clerk's and the bag-boy's diary recollections of the event?

It doesn't take much research in a University Library to realize that the
history of a place, and of a people in that place requires searching beyond
University collections, and deep searches into the text of documents. No
one cultural institution holds everything that we might want to reference or

In fact, Ephemeral Cities recognizes that the history it seeks to uncover, in
many cases, is not held by any cultural institution. So, it lends an ear to
the community, and calls out: "Become a part of history !" "Make history,
your story !" Or, to borrow a turn of phrase from Monty Python: "Bring out
your dead !" Remember, the man in the cart is not quite dead yet.

final project report page 66


I like to tell the story of Mary Todd Lincoln. In 1868, Mrs. Lincoln's maid,
Elizabeth Keckley wrote a tell-all tale of life in the Lincoln Whitehouse.
Appropriately entitled, Behind the Scenes, it is virtually the only such
published account. By contrast, a contemporary presidency will generate
at least two purportedly intimate accounts. And, recent presidencies have
generated far more than that!

The title among those digitized by the University of North Carolina -
Chapel Hill exposes a vindictive, mean-spirited woman. History
remembers her so, rather than as the tragic figure later cut for Mrs.

Obscure, even today, among the archival collections of Transylvania
University, alma mater of Todd men, letters of a younger Mrs. Lincoln
under-pin the rage of a woman all but abandoned. They lend reasons to
Mrs. Lincoln's apparent irrational fits of anger. And rationale for
partnerships among archives, libraries and museums.

One assumes that the price of milk will remain a constant of historical
price comparisons, a surrogate for the economic health of the nation. But,
who will long remember, much less understand, the importance of a
barcode scanner in presidential politics without museums preserving and
interpreting the artifact.

Ephemeral Cities postulates that uncovering this hidden history can lead
to a sea-change in our understanding of history.

final project report page 67

History, for too long, has been capitalized rarely socialized. Politics
aside It really does "take a village", as Hillary Rodham Clinton put it, to
revive a community's long past sense of itself.

More importantly, Ephemeral Cities Suggests that, by bringing together a
whole communities' resources, we can SEE change. And, if we can see
it, we can understand it.

So, here's a graphic example.
Ephemeral Cities will repurpose
the historic SanbornRR Fire
Insurance Maps of Florida,
making use of their wealth of
detail and accuracy. Scanned
earlier, and digitally restored to
their as published state, we've
subsequently geo-rectified the
maps, so we could use known
earth coordinates. As a means
of laying one map atop another.

Here, new uses changed the
character of Tampa. This block
near the waterfront went from
m in shopping district to after-hours
entertainment center, as
transportation took more goods
inland and as more goods took
more men to handle and deliver
them. Following Geo-
rectification, buildings are
indexed both graphically and

^B ^Ljj-J -L- i .,J JY^ ~ ^ LJ --L i? twi&k-r^uL&

final project report

i'-. 1900


Jk "
J.l.- .wInty.

page 68

Graphical indexing placement
of the red dot associates
relatively precise longitude and
latitude with the building.
Textual indexing records street
location, building number and
name and any other recorded

Textual indexing becomes the
base of a gazetteer or place-
name authority file that will be
augmented subsequently with
information from city
directories, land ownership
files, and other information
sources, all to facilitate

So, here's the concept in action. (N.B. This is still a demonstration,
representative of the web-site still under construction.) The user has
zoomed into a location of interest on a period Sanborn map : here,
Gainesville, 1884. The user may either point and click, or, construct a
particular search strategy; here, by building Information.

IllU* ~ ~ -- -I Y
, ~IYYI l ~i ll 'Ir Ii rl 1 I L I ...
Ill ........... ... Iii i ii


I ,. ,r ,,,- .,

^B ^Ljj-J -L- i .,J JY^ ~ ^ LJ --L i? twi&k-r^uL&

final project report

page 69

In this case, the

user has decided to search "Building Use" for "Cigar

The cigar factory in this block is identified by a
desired, all of the cigar factories in Gainesville,
target cities could be seen.

red dot. If the user so
or, in any or all of the

Clicking on a red dot (or, on any building) displays all of the known
information about that building's use and occupants at that approximate
time in history. Advanced queries will allow searches of building uses and
occupants over time. As we've already seen, this information is extracted
from name-rich resources.

^B ^Ljj-J -L- i .,J JY^ ~ ^ LJ --L i? twi&k-r^uL&

final project report

When the user clicks on an alternate use say, Grocery red dots
indicate the location of other grocers. My favorite proximity study is that of
schools to churches to saloons and their impact on public morality.

By clicking the red dot of any of these locations, information about that
location is displayed. Clicking either a use or an occupant's name
launches a query against targeted collections. Retrieval lists, sorted by
holding institution, display a thumbnail together with brief descriptive
information. A selection invokes a new browser window to display the
selected resource, within the holding library's web space.

Let's briefly look into the future. If we assume that the society of the
future, as it is today, is mobile. We can assume that information itself will
be mobile. Here, I envision a future of aware personal devices that deliver
information in what I'll call "touring mode." In this future, time travel, in a
sense becomes possible.

page 70

final project report page 71

This information collection strategy, by the way, works well for a small
community of targeted web-sites but does not scale-up well. We hope to
see our other methods the concept of geo-temporal searching, for
example, adopted by what are known as OAl Harvesters, such as
AmericanSouth.ORG. A harvester is a tool that, like Google, goes out and
collects this information in advance. But for the moment, harvesters
collect only information about Internet resources. They don't collect or
mirror or duplicate the actual resource as they would need to, to find
information from within document texts.

Ephemeral Cities partnerships are structured as originally envisioned by
PALMM (the Florida digital cooperative) in hub and spoke fashion.
Expensive and highly skilled activity is centralized; all else is distributed.

The hubs of this pilot project are Gainesville, in north Florida; Key West, in
south Florida, and Tampa, in mid-Florida. These cities were chosen, first,
for their willingness to participate in a project with several significant
technical challenges; ... as for the depth of their local-history collections,

Hub &

final project report page 72

each representing a different state of Florida; ... and for the strength of
their digitization programs.

The Gainesville partnership brings together several strong local-history
collections, all but one with a history of digitization. The Key West
partnership likewise brings together strong local-history collections, and
again all but one with a history of digitization. The Tampa partnership
brings together strong local-history AND museum collections, as well as a
demonstrated commitment to distance learning.

The project also draws on the PALMM collections and the contributions of
institutions across the state.

We're confident of success. And, many of our partners outside this project
have begun digitizing local- and regional histories.


final project report page 73

Now, let's look at variations on the hub and spokes. This model of
imaging is typical of PALMM. Small institutions scan; feed to hubs that
manage the images created, which feed to a central archive. This model
emphasizes preservation while, at the same time, moving more complex
and more expensive tasks up the line.

This text processing model demonstrates additional centralization. All
page images pictures of pages are sent to the University of Florida,
which will return searchable text versions to those web servers capable of
searching document texts. In this case, the Prime Recognition software
being used for text conversion is extremely expensive to license and

The query model demonstrates, by
far, the most centralization. This
Ht model gives the Internet user one-
-, stop-shopping, as it were.

The EpC [Epoc] Server does the
..-, labor of searching multiple sites and
w U returns a unified list of resources to
S.---- the user.

Quickly, targeted materials include: archives, monographs and pamphlets,
-- which, I should note, include oral history transcripts; maps and plans,
photographs and postcards, serials, and especially, newspapers, and,
finally, artifacts and specimens.
finally, artifacts and specimens.



final project report

Regarding specimens: You may not know that train transportation was
King in Florida during this Project's target period. A motto of Florida's
economy could then and might still be: "Tourists in! Vegetables out!"

Having seen the demo, some of you are probably wondering "Isn't
there tons of labor involved?" "How do they do that?" I have the proto-
typically Florida answer:
Mermaids! For those of you
who recognize the second
question: "How do they do
that?" from the Britta TV ad
featuring children, you know
that there are boat-loads of
unemployed mermaids.
Specifically, the mermaids
known as: imaging automation,
text conversion, parsing, and
name authority.

page 74

t b_* --

^B ^Ljj-J -L- i .,J JY^ ~ ^ LJ --L i? twi&k-r^uL&

final project report

New methods and programming will arise from the project's murky deep
and we hope will be adopted in other states to build the national map.
... 1, ;: i : ;: ;

Our faith in automation is

We've made

the following

* Everything can be fixed in
place and time, and,
* Semantic Processes can
be applied to automate
this work

I should make clear, here, that our users are (anticipated to be) historians,
genealogists and students, ... but also naturalists, real estate agents, land
developers and certainly, not last and not least, tourists on "voyages out of
the ordinary!"

If you would like more
Cite (information on the technologies
we employ, these links should
be helpful. You can also find
an elaborated workflow our
text-processing applications on
our web site.

The Florida partners invite you to join us in reviving other Ephemeral
Cities (http://www.uflib.ufl.edu/digital/collections/EphemeralCities/).

Versions of this presentation are available online in power point and PDF
in English, French and Spanish at http://www.uflib.ufl.edu/digital/

page 75

- EveryiNing

- Semantic Prollesses


final project report page 76

Appendix B : Authority Records
for Information Extraction and Query
Mark Sullivan & Erich Kesse

Think of Ephemeral Cities as a kind of multi-act play, an epic that tells the
story of a people. Though our story covers only forty years, during the
dawn of the 20th Century, it has a cast of thousands. There are thousands
of people and thousands of buildings among thousands of other named
things and events for each of the project's three targeted cities.

Each required disambiguation: father from son, mother from daughter from
distant cousin to individuals of no relation whatsoever other than shared
name. For me, the issue is personal: for the first twenty years of my life I
shared my name, same spelling, with a cousin twice removed, born on the
same date in the same hospital, who attended the same schools, before I
moved away. Imagine the same problem in a more closely knit
community, one that valued heritage by handing names common as
common down. Disambiguation was also required for buildings and
businesses. A corporate name could rise and fall, lay dormant, then rise
again. Or, a name could be shared, one with another in a distant cities,
sometimes even the same. At the outset, it should be said that Ephemeral
Cities greatly under-estimated the tangle of history. We needed a record
structure that could be easily augmented as new data was extracted but
which was sufficiently delimited to minimize data reprocessing.

THE PROBLEM? There are almost four million "First Baptist" churches in
the U.S.A. today. There are at least another 240,000 "Second Baptist",
but only 22,000 or so "Third Baptist". Incredibly, there are only a few
hundred Starbucks cafes in New York City, of which 44 were reportedly
cited by health inspectors for one violation or another, any one of which
could have been a home-away-from-home for any one of the City's 112
John Smiths as he writes the next great American novel. Think of the poor
historian of the future who has to sort that mess out. The problems faced
by Ephemeral Cities paled by comparison, but with records spread across
repositories, street addresses without numbers, and with names
appearing in records without sufficient disambiguating data or, worse, with
only the most obtuse of clues, the problems may have been as vivid.

^BB^J-jjI-JL -I -L i .1, l lY A l lJ J BlL i? itliik&r uL&

final project report

This Appendix outlines the Authority Record System constructed for
Ephemeral Cities. This is not a full discussion of rationale or of choices
made. More discussion is planned for an article to follow in publication.

The outline of the Authority Record System for Ephemeral Cities, now
being implemented for all of UF Digital Collections, is found in this
zoomable chart, best viewed in Internet Explorer, at http://www.uflib.
ufl.edu/digital/development/authority/Authority files/Authority frames.htm
or in large image view at http://www.uflib.ufl.edu/digital/development/
authority/authority svstem2.iDa.

Feature Infomrttion




Gieo-~SpatiaI Infonuation



Figure B1....... Ephemeral Cities Authority System: a simplified overview of major information categories.

The system has five categories of information: Geo-Spatial, Feature,
Street, Person and Corporate information (Fig. B1). It lacks a sixth
category currently under consideration: Named Event Information. With
few exceptions, this system also lacks linguistic support; the rationale is
simple enough: a translation layer can be developed separately and

page 77


final project report page 78

placed atop any given node or element later. Such a translation layer is
currently under development for use in our U.S. Dept. of Education TICFIA
grant funding the Digital Library of the Caribbean.54 Such a layer was not
immediately necessary for the Ephemeral Cities project.

Most features-based authority systems, e.g., that of the Alexandria Digital
Library, recognize streets as a feature type. It is called out separately
here because it is a relative complex type for Ephemeral Cities. Streets,
themselves, have features. And, streets, street names and street features
change over time. Handling of Street Information is the most obvious
indicator that the Ephemeral Cities Authority System's structure has been
designed both for the information dealt with and the data mining and
information extraction behaviors that we expect.

BACKGROUND IN BRIEF. Traditional MARC authority records55 are far to
narrow. They relate names and provide referencing information
accommodating dates and relationships; but, they don't make this
information searchable in a meaningful way, let alone with the nuances
needed to facilitate information extraction, machine aided disambiguation,
and discovery.

Evolving standards for authority within the archival records community
delimit nuances. It's Encoded Archival Context (EAC),56 an
implementation of albeit, a variation from the International Standard
Archival Authority Record (ISAAR),57 provides a fuller record. But, neither
EAC nor ISAAR is fully delimited (or sufficiently relational), as yet, for use
in information extraction. Work ongoing with North Carolina ECHO's
North Carolina Encoded Archival Context Project58 is designed to facilitate
creation of EAC records in machine-aided production environment.
However, it has not yet been awakened to, re-purposed for automated

54 See: http://www.dloc.com/
55 See: http://www.loc.gov/marc/authority/
56 See: (a) Yale University: http://www.library.yale.edu/eac/, (b) the University of Virginia:
http://www.iath.virginia.edu/eac/ and, for the most recent work, (c) North Carolina
ECHO's North Carolina Encoded Archival Context Project: http://www.ncecho.org/nceac/.
57 See: 2nd Edition as adopted (October 2003): http://www.icacds.org.uk/enq/ISAAR(CPF)
2ed.pdf or as finalized for publication (2004): http://www.ica.orq/sites/default/files/
58 See: http://www.ncecho.org/nceac/; see also, North Carolina Biographical and Historical
Information Online (NCBIO) at http://www.ils.unc.edu/digitalnc/ncbhio/index.htm


final project report page 79

information extraction as a production tool and not yet optimized for post-
production use in discovery.

Both MARC and EAC (as well as ISAAR and authority record constructs
within TEl and other schemes) are constructed largely for human use and
limited computerized uses. A typical authority record, however composed,
wherever it resides, is typically designed for a publication format, intended
to be read by human eyes, and to be used for disambiguation without
machine-assistance. Records are lightly tagged and most known
information is recorded in narrative strings. This is not to say that these
records cannot be used to disambiguate persons and places; they can.
But, revision of these standards, or creation of a wholly new standard for
use by code, in programming, is needed to facilitate machine-assisted
disambiguation. Such action will build more complex, even heavily tagged
records to be certain. But, these more complex records will not allow
known information to go to waste, unexposed to automated semantic

Relative to place names, the U.S. Board on Geographic Names (US BGN
or BGN), maintains the most comprehensive authority records more
than 2.5 million names for the United States of America. Its Geographic
Names Information System (GNIS)59 provide name information,
geographic positioning data, coding references, feature characterization,
etc. GNIS provides some support for alternate and historic names, but
little delimitation of historic relations. And, while the GNIS database is
extremely large, it often lacks local and particularly ephemeral names.

GEO-SPATIAL INFORMATION (Fig. B2). Geographic information begins with
the data types typically found in the USGS Geographic Names Information
System (GNIS). Ephemeral Cities concentrated effort building authorities
in this area in order to strengthen place associations. The central table
establishes a "RegionlD", identifying a name, associating it with a type of
region and a parent region and with a Federal Information Processing
Standards (FIPS) code. FIPS coding ensures that Ephemeral Cities data
can be associated with external sources, particularly with U.S. government
information such as U.S. Census reports but also with local data. Other
associative data, establishes geographic bounds and historical notes.
Geographical bounds, footprints, fix either a point or box (sometimes a
polygon) for a given place. Footprints, also inscribe with temporal data,

59 See: http://qeonames.usgs.qov/domestic/index.html


final project report page 80

are a way of saying, "This place existed here at this time." Other history
notes, allow authority systems for fix names, without footprints, between
start and end dates; to associate a name with older names, also with start
and end dates, as well as with alternate forms of names and both their
linguistic and temporal goal posts of usage of the alternate forms. This is
a very rich means of establishing geographic authority.

Some geographic authority information is given a separate and somewhat
special place. Street information, for example, is set aside to ensure
proper relations between more than one geographic region. The locations
occupied by individuals is also given similar special place, again to ensure
that any one individual can be associated with the places that they lived,
worked, played, died, etc.

Figure B2. ...... Geo-Spatial Information.


final project report page 81

STREET INFORMATION (Fig. B3), again is a special reserve of geographic
information in that it may appear in more than one geographic parent.
But, street information also changes over time. Authority information
establishes name, provides a unique identifier a common function of the
system and of most relational databases, as well as start and end years.
Street data functions in special ways. The fact that street names may be
associated with more than one geographic area think of Route 66, "from
Chicago to LA" has already been noted. The function of postal
addresses and the association of addresses with people, corporate bodies
and events further enrich the authority data maintained.

Figure B3. ...... Street Information.

FEATURE INFORMATION (Fig. B4) is another special reserve of geographic
information. Feature information is a way of characterizing location, e.g.,
"This is flat land"; "This is an island"; or something more esoteric: "This

^BB^JL-jjI- JL -IL -L iV. ~ ~~1111111&11111 L L -B.L i? tlti Sr L&

final project report

page 82

data is a postal address." Features
longitude; latitude; Albers coordinates;
employed as a kind of short hand;
associates it a specific type of feature.

Feature Information

Three are different typer of K -
ULFODCF F8ltu TI (fatUfre F ,- TAi.
(mountain range, building 1 ) ..

j: i l.. J UL.UI Fealuis AuLnCuJe
F 1 Fe dlurlty llO

UFDCFea I-eU-e A realure may have -FDC_F-J I e_UWLJr
PK Fean driffearnl uses over tim PK FUMji
PK.4 FalmPK aKlD
FK1 PaenlFnatureUsPlD lln1 "r
FIterval E ea

bear a name; location description;
and altitude. A feature type code is
a record that contains the code

A feature appears wllhin on
or more geographic regions
UFDC_FFea urm_R.Fgt .Link
PK.FK1 Feaiel

A feature may be usd
by a person as a home
for some period of time

Features may appear on street

During a year rang, a
feature may function as
a business location

A feature may be ussd
by a company during a
certain year range

Figure B4....... Feature Information.

Historical notes and temporal goal posts are also established. It is
possible to record earlier and later dates, changing business uses and
changing corporate names, as well as to associate a feature or features
with given geographic regions, person, corporate bodies, and uses for
specific and changing date ranges.

PERSON INFORMATION (Fig. B5) can be especially detailed and complex.
The Ephemeral Cities authority record structure attempts to deal only with
constructs useful for disambiguation for persons living between 1880 and
1920 in the Florida target cities. It needs to be adapted for a more


final project report page 83

complete and complex set of data relationships. Life events are under
represented. A plan for authority record enhancements has been
specified but is not yet in place.

The central table of the current data structure establishes a Person ID that
can be used as short-hand for the group of data elements including name,
parsed into first, middle and last, and associated with a given title, sex and
race. One weakness of the data structure here is its failure to
accommodate changing titles most often through marriage, e.g., Ms. to
Mrs and sex, M-F or F-M; cases of the latter were not seen in
Ephemeral Cities but certainly become an issue for the present.

Figure B5. ...... Person Information.

Inasmuch as possible data elements used here are fixed to the U.S.
Census. This, of course, beyond the sensitive issue of classifying race,
leads to issues of dealing with changing descriptors over time. Race is
handled as reported; no attempt was made to apply a filter or politically
correct revision. No data changes more over time than historical
occupation. Here, occupation, like race, was recorded as reported,
leaving the weightier issue of flattening the various census descriptors for
a later date.


final project report page 84

The record of temporal start and end, born and died, markers for persons
is weak; redundancies built into the Geographic Information are excluded
here not intentionally, rather as an indicator of development as we
attempted to deal with an overwhelming flow of data. An assumption,
working from within the geographic bias of Ephemeral Cities, is made in
the current iteration of the data structures that temporal markers can be
found in location information (i.e., in the Person_Home_Location and
Person_Region_Link nodes). This, of course, presumes that we know
where an individual was at all or at least at the start and end of one's life.
The presumption, based in the certainty that digitizing and parsing
directories and the like fixes a full base layer of person information, often
proved to be false.

While we concerned ourselves with uncovering the history of the common
person, the authority structure suggests a preoccupation with the common
man. Work-life (e.g., historic occupations and work locations), with
authority record temporal markers, for the date range targeted by
Ephemeral Cities details the lives of men more than those of women. The
greater fault lies with the documents from which data was being extracted;
they centered in this pre-universal suffrage era on the adult male. But,
the fault rests also with the project design, which might have done more to
reflect a broader range of life events: marriage, birth, baptism, etc. We
have since specified revisions enriching the authority system with a full
range of life event tracking, but have not been able to implement it as yet.

CORPORATE INFORMATION (Fig. B6), after Geographic Information, received
the most attention in the Ephemeral Cities project. The reason goes back
to the Sanborn maps. The livelihood of businesses takes the form of
names: the names of hotels, the names of businesses, even the names of
stables on the maps. The lives of persons, however, is summed in one
nearly universal label: "dwelling". Because the modern concept of postal
address wasn't established for decades following the dates targeted by
Ephemeral Cities, dwellings with distinctive footprints have ambiguous
printed references in the documents scanned for Ephemeral Cities. For
example, who's to say which house "John Smith, University Avenue" lived
in or how many of his neighbors had their mail, their newspaper
announcements, and their legal documents addressed "University Avenue
near the Corner of Main Street"? ... let alone which corner: NE, SE, SW,
NW or how near the corner: one or two or more houses from the corner?


final project report page 85

Corporate information records a name, assigned an ID, with a descriptive
note (here, labeled: Memo the Authority Systems data elements are in
need of some consistency in labeling), and associated with a Parent
name. Corporation records link out to historical notes and occupations,
with their own temporal start and end markers. Lacking is a method to
designate businesses that moved. But for business failures, the date
range 1880 through 1920 targeted by Ephemeral Cities was stable.
Businesses didn't tend to move; they were born and died in one location.
So, the considerations given streets and persons are not given to
corporations. An enriched structure, optimized for a greater date range,
including the contemporary period, will have to accommodate moving

A feature may be used
by a company during a
certain year range

UFC _H.slorpaLoKk paa A person Is employed by a
CorpPK HiltoealO- ultienlD company for a particular
Corporate -, job at some location
Sn fr\ rm a n n iHistorFccupon du during a year range
Inform ationI Ieoi
/Oup-atonlD UFDCPersaon_0oJal.oLin
UFDC_Crporfale_Bod, PK aPermnCurLnklD
PK CtolraWlI FK3 PersnIlD
SFK1 CorporatelO
CorporaeNamt m FK2 HistoricalOccupaUonlD
Memo A company employs stanYear
U2.UI CorpAuthCode people to perform Jobs EndYear
FK1 PafsenCorpoalelD Woklocatlio
FK4 Wo~kFealurelD

Figure B6 ....... Corporate Information.

CONSTRUCTION OF AUTHORITY RECORDS is for the most part, initially a
greatly labor intensive process. With time, as records are enriched and
queries against searchable text using information extraction programs are
run, the process grows more automated. This was the theory under which
the Ephemeral Cities project was conceived. Project planners continue to
stand by the theory. But, the initial stage was far more labor intensive
than the Ephemeral Cities project planners had envisioned. Very little
automation gain was realized. This weakness of the project can be seen
in the frequent failures of the query systems to connect geo-referenced
maps with text and images in the digital library.


final project report page 86

Appendix C : An Alternate Interface
A Google Earth Experiment Key West
Laurie N. Taylor & Erich Kesse

Google Earth was not available at the time that Ephemeral Cities was
conceived, planned or funded. It has since become a viable means of
information delivery, capable of wrapping maps and information points into
a single interface. It also remains, however, as the product of a
commercial body, outside the sphere of public ownership.


Several of the Sanborn Fire Insurance Company maps for Key West from
1889 and 1899 were added to Google Earth to test usability. Additionally,
several archival photographs, digitized by Ephemeral Cities, where then
linked to the maps. A more active entry to the content of Ephemeral
Cities, i.e., database queries fashioned as URLs, were not added for
purpose of this demonstration; addition is planned for subsequent live
versions of Ephemeral Cities on Google Earth. Google Earth afforded
major improvement in ease of use over the Ephemeral Cities ArclMS

While ArclMS offers an exacting environment for spatial exploration, its
visual interface is difficult to navigate, even simplified as it is in the
Ephemeral Cities GIS interface. Unlike ArclMS, Google Earth offers an
easy interface for novice users. They can readily see and grasp controls,
as well as, relationships among the displayed elements. More importantly,
the material displayed by Google Earth is encapsulated within a portable
KML standard format. KML, or Keyhole Markup Language,60 is an XML
file format for managing the display of three-dimensional geospatial data.
While the KML standard was developed for Google Earth, the KML
version 2.2 specification has been submitted to the Open Geospatial

60 Keyhole Markup Language: see, http://en.wikipedia.org/wiki/Keyhole Markup Language


final project report page 87

Consortium61 so it can be established as an open standard for all geo-
browsers. Because KML can already be used with other applications and
because it is in the process of becoming an open standard, representing
materials within KML allows those materials to be used in a number of
different programs. Thus, while the use of KML was investigated through
Google Earth, the files created for use in Google Earth could be used
within many other applications ensuring sustainability and interoperability
with other projects.62

Figure C1....... Google Earth: Basic interface showing easy visual display of multiple layers in Google Earth
for Key West. Sanborn Maps and historic photos (as placemarks) are shown.

61 Open Geospatial Consortium: see, http://en.wikipedia.org/wiki/Open Geospatial
62 Applications currently offering some support for KML include: ESRI ArcGIS Explorer
(http://www.esri.com/software/arcgis/explorer/), NASA World Wind (http://worldwind.arc.
nasa.gov/), and Google's SketchUp (http://sketchup.com/), which is being used to provide
a 3D experience to explorations of maps and building spaces.


final project report page 88


The image above (Fig. C1) shows the ease of use in representing
multiple, layered types of data within the Google Earth interface. The
main screen has side-bar tools, to the left of the screen, including search,
specified place information, and layers. The rest of the screen is covered
by the map interface.

The presentation, for this demonstration, is relatively simple. For
simplicity, we've turned on all of the maps and specified places known in
Google Earth speak as placemarks (sometimes referred to as placepoints)
that populate this demonstration. Future packaging is likely to nest
content, both maps and placemarks, for any given city, chronologically in
layers that can be turned on and off individually, allowing the user to
explore only within given times of personal choosing. The data for any
given city will be available separately, allowing them to be added, turned
on or off, and searched in modular fashion: as much or as little as they
user prefers.

For purposes of this demonstration, the Sanborn maps are nested within
the "Places" area of the side-bar. Each map can be turned on or off and
made transparent using the slide-bar located between the "Places" and
"Layers" areas. With all Ephemeral Cities content selected (i.e., checked
or ticked on), the mapped area shows placemarks for Ephemeral Cities
photographs. Placemarks offer an easy visual icon to show where the
images are in relation to Key West landmarks, past and present. Further,
the titles on the maps and the photographs provide text which can be
searched using the top "Search" tool from the left-side tool area.

Google Earth allows GeoTiff files to be imported using the geographical
metadata encoded in the files, and for files listing sets of placepoints and
their information to be imported. Rather than having to write KML files for
the placemarks and maps, KML files can be created by importing the data
into Google Earth and then returning KML files. KML files can also be
created without Google Earth and then displayed within Google Earth or
any other KML browser. This ease of file creation and ease of use
supports the vision of the Ephemeral Cities project a visual, community
atlas that users can explore and contribute to.

One disadvantage of this approach is that the community KML must be
periodically and actively rebuilt. It does not change dynamically. We

^BB^J-jjI-JL -I -L i .1, l lY A l lJ J BlL i? itliik&r uL&

final project report

page 89

continue to investigate the use of automation to generate the community
KML over time. Another disadvantage of this approach is the requirement
that the user download, then click the KML to view and use it.

Figure C2a..... Key West, 1889 map sheet overlay illustrating use of transparency

Figure C2b.
Overlay with street names from both map sheet and
Google Earth identifies a street name change or
Note that match is not exact, the artifacts of both
error in the map sheet and imprecision of non-
military satellite imagery in Google Earth

Figure C2c.
While the artifacts of inexact match can be seen
clearly, changes in building density and extent of
architectural preservation since 1889 can also be
Map sheets covering the same area over time could
have been layered atop one another to reveal
change overtime.

The screenshots that follow (Fig. C3 through C7) show some of the data
imported and created in Google Earth. We imported GeoTiff files from the
Ephemeral Cities project, but the outsourced company that had created
the GeoTiff files had not made the blank sides transparent as this was
unnecessary for the ArclMS interface (Fig. 2Ca). Because the sheets
have some degree of overlap and because those areas were not
transparent, merging the sheets within the Google Earth interface proved


final project report page 90

difficult (Fig. 2Cb-c). Further, the changes to the streets and inherent
inaccuracy of the historical maps (especially for outlying border areas)
made them less usable within Google Earth's interface. Given these
issues, we chose to import lower quality images and then edit them within
Google Earth so that the maps could be as usable as possible in relation
to the changes in the city's development over time. Choosing lower
quality images was also useful for ensuring faster loading times because
the images must be supported on an external server that Google Earth
references and pulls the images from for display. We matched the images
to the current street locations (based on the "roads" layer in Google Earth;
see, Fig. C3).

Nntasae ad pa 1t s swh alaye

Figure C3. Non-transparent and partially transparent sheets with "road" layer.
Figure C3. Non-transparent and partially transparent sheets with "road" layer.

Once the Sanborn maps were overlayed onto Google Earth, placemarks
were imported from the Ephemeral Cities database containing information
extracted both from the maps and from verified information sources
digitized by the project (Fig. C4). All placemarks, shown with their labels
in the main Google Earth interface, have textual titles. Google Earth also
makes possible an association of placemarks with images that can be
displayed along with or in place of the text (example not illustrated). Any


final project report page 91

placemarker may be clicked to reveal details about that place. Details
may take the form either of images (Fig. C5) or narrative (Fig. C6). Image
and narrative may also be combined. And, details may include an
embedded query (Fig. C7).

,:,, Googl i E i',Prol

Fig LaC. P lacemars imprtd from a reads .
Figure C4. Placemarks imported from a spreadsheet.

Queries built for this demonstration, searched the UF Digital Collections,
including the Ephemeral Cities database, Ephemeral Cities photograph
collection, and other textual, visual, and audio collections. For all of the
advantages and strengths of this approach, of the Google Earth interface,
queries continue to be a weak point. As in the Ephemeral Cities GIS
interface, queries are reliant on the strength of the authority files.

Google Earth presents another new weakness. The user has to find a
given location. Google Earth makes the assumption that the average user
knows something about the space in which they are navigating.
Ephemeral Cities, in work with patrons and school children, is unable to
make that assumption. To compensate somewhat, we enrich the KML
inasmuch as possible to increase textual-search success within the
Google Earth Interface.

^BB^J-jjI-JL -I -L i .1, l lY A l lJ J BlL i? itliik&r uL&

final project report

IFle &t yew IMcs Arl tiep

Figure C5. Photographic details of a given place.

Sanhorn Insurance Maps, 1889
These are the 'Saborn@ Fire Insurance Company Maps of Florda' fr
blblographic units in more than 3,000 map sheets

bridges, docks and barns Along wth fire statons, you could also fin

railroad nes adnd ian reservatns and topography were included
Shisory, land use, and hioric preservation

4., p p _

d water fracliles, sprinklers hydrants, csterns, and alarm
.. .
Today, the mapsare n invaluable gide to inner y

L4L 6 4. Ar*

PtvrI~ir.&aJ M-j bil &iE4EAC .nbi; sw

Figure C6. Textual details of a given place. (Displayed here, the textual detail pertaining to a
map sheet as a whole. Details, as easily, could have been of a specific place.)

page 92

IGgLe Erh r 1-r][X



final project report


Figure C7. This shows the detailed information for one of the placemarks.

The Ephemeral Cities database and the digital collections supporting the
project are much richer and dynamic than the information that can be
presented in the KML file. Historic notes, alternate names, occupants,
building uses, and other information is added to the KML's descriptive
notes, those either viewable or hidden, for a given place.

Construction of Google Earth access points (i.e., community KMLs) is a
response to the failures of the Ephemeral Cities (ArclMS) GIS interface.
Construction and testing of various approaches is ongoing. Some facets,
some of the working assumptions made to support ArclMS, such as the
transparency issue, must be readdressed for use in Google Earth. The
transparency issue, for example, requires reprocessing the maps to
remove the background white. That task will be time consuming for the
194 map sheets that comprise the three current Ephemeral Cities
(Gainesville, Key West, and Tampa). They certainly will not be trivial as
we reprocess the 3000+ maps that represent each of the one hundred
Florida cities that we have hope to expand the project to include.

page 93

t b-*- --

G..l. .,t Pr F FM1 O

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs