Integration of decision systems with production information for operations management

MISSING IMAGE

Material Information

Title:
Integration of decision systems with production information for operations management
Physical Description:
x, 149 leaves : ill. ; 29 cm.
Language:
English
Creator:
Khuri, Ramzi S., 1964-
Publication Date:

Subjects

Subjects / Keywords:
Citrus fruit industry -- Costs -- Databases -- Florida   ( lcsh )
Citrus fruit industry -- Management -- Florida   ( lcsh )
Citrus fruit industry -- Cost of operation -- Florida   ( lcsh )
Agricultural Engineering thesis Ph. D
Dissertations, Academic -- Agricultural Engineering -- UF
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1990.
Bibliography:
Includes bibliographical references (leaves 143-147)
Statement of Responsibility:
by Ramzi S. Khuri.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001687942
oclc - 25116703
notis - AHZ9972
sobekcm - AA00004762_00001
System ID:
AA00004762:00001

Full Text











INTEGRATION OF DECISION SYSTEMS WITH
PRODUCTION INFORMATION FOR OPERATIONS MANAGEMENT









BY

RAMZI S. KHURI


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


1990


































Copyright 1990
by
Ramzi S. Khuri
































Dedicated to
my mother Maha, my father Suhail,
my brother Nizar, my uncle Zahi,
and my best friend Wisky














ACKNOWLEDGEMENTS


I would like to express my deep hearted gratitude to Dr.

R. M. Peart, Graduate Research Professor of Agricultural

Engineering, for serving as a chairman for my graduate

committee. I would also like to extend my appreciation to Dr.

Howard Beck of the Department of Agricultural Engineering for

his knowledge, guidance, and supervision throughout this

project. The friendship, support, and guidance of Dr. W. D.

Shoup of the Department of Agricultural Engineering, his help

in formalizing the project, and his serving as cochairman on

my supervisory committee are highly appreciated. I extend a

special thank you to all those who were involved with the

continued financial support I received for this project. I

would also like to extend my gratitude to Dr. Richard Kilmer

of the Department of Food and Resource Economics and Dr. James

Burns of the Department of Industrial and Systems Engineering

for serving on my committee and for their help on this

project.

Mr. Ron Muraro and Dr. Megh Singh of the Citrus Research

and Education Center in Lake Alfred, Florida, deserve my

special thanks for sparing their time and providing me with

valuable information. I also thank all participants of the

cost surveys, and those who helped evaluate the COINS system.

iv








I would like to express very special appreciation and

thanks to my best friend Betsy, whose patience, encouragement,

friendship, and moral support were indispensable during the

completion of this project. I would also like to thank

Betsy's family for their friendship and continuous

encouragement throughout my studies.

My very special thanks are extended to my uncle Waleed,

my uncle Rabie and his family, and my best friends Sami and

Ghassan for their endless support during my five and a half

years of graduate study. My appreciation is also extended to

all my friends in the United States and abroad.

Last but not least, I wish to express my appreciation to

my grandparents whose confidence in me and continuous support

for my efforts helped me achieve my goals.















TABLE OF CONTENTS
pages
ACKNOWLEDGEMENTS ...................................... iv

ABSTRACT .......................................... ....... ix

CHAPTER

I INTRODUCTION ................................. 1

Justification................................ 1
Objectives................................... 3

II REVIEW OF LITERATURE......................... 6

Management Information Systems............. 6
Agricultural Information Systems......... 7
Agricultural Technology Transfer.......... 8
Agricultural Cost Systems................ 9
Decision Support Systems..................... 11
Expert Systems........................... 13
Expert System Applications ............... 14
Database Management Systems.................. 17
Database Design ................... .... 17
Data Security ........................... 19
Integration of Knowledge Systems
and Databases ................... ....... 20
Summary..... ................................. 23

III SYSTEM DESIGN CONSIDERATIONS................ 24

Preliminary Research........................ 24
Existing Information Sources............. 24
Existing Agricultural Software............ 25
Design Requirements ....................... 26
The User Interface ........................ 26
System Hardware and Software
Requirements ............................. 27
Functional Requirements........ ........ 28

IV COINS CITRUS COST INFORMATION SYSTEM ....... 29

An Overview of COINS ................... ...... 29
DBXL as a Database Management and
a Programming Environment............... 30









pages


Program Development Tools................. 30
Program Requirements ..................... 31
Program Structure......................... 31
Password System .......................... 32
The COINS Database............................ 32
Database Structure........................ 32
Data Storage .............................. 34
Data Entry Program............... ........ 38
Program Design.......................... 38
Program Functions......................... 38
Entering Costs for a Grove................ 47
Data Summary and Comparison Program.......... 53
Program Design........................... 53
Program Functions......................... 55
Cost Averaging............................ 55

V SIMON SYSTEMS FOR INTEGRATING MANAGEMENT
AND COST INFORMATION..................... 59

Program Development........................ 60
Program Features............................... 61
Program Development Environment........... 62
Expert System Environment................. 62
Expert System Rule Base................... 65
Knowledge Acquisition........................ 67
Identifying the Experts ................. 67
Preliminary Interview.................... 69
Sources of Information................... 70
Establishing the Decision Process ........ 71
Follow-up Interviews..................... 84
Program Execution .......................... 84
Obtaining a Herbicide Recommendation...... 84
Obtaining a List of Weeds Controlled
by a Herbicide............................ 90
Comparing Several Herbicides ............. 90
Integration with COINS .................... 92

VI TESTING AND EVALUATION......................... 96

Phase 1 Operational Evaluation............. 96
Testing and Evaluating the COINS
Cost Programs ............... .......... 96
Testing and Evaluation of the
Herbicide and Weed Guide ............... 99
Phase 2 Professional Evaluation ........... 100
Role of Trade Show Exhibit............... 100
Role of Citrus and Extension
Specialists ........................... 102
Purpose of Evaluation Form .............. 103
Evaluation Procedure..................... 103


vii











Evaluation Results for
Structured Responses...................
Evaluation Results for
Non-structured Responses................

VII CONCLUSIONS AND RECOMMENDATIONS..............

APPENDICES

A DATABASE FILE FORMATS ........................
B SYSTEM EVALUATION. ............. ..............

REFERENCES ... .. ..... ....... ..... ... ....... .. .......

BIOGRAPHICAL SKETCH... ............ .............. ......


viii


pages


105

110

122



126
129

143

148














Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

INTEGRATION OF DECISION SYSTEMS WITH
PRODUCTION INFORMATION FOR OPERATIONS MANAGEMENT

by

RAMZI KHURI

December 1990

Chairman: Dr. R. M. Peart
Major Department: Agricultural Engineering


Production costs for citrus operations in Florida have

been integrated with extension and management recommendations.

This new approach permitted the dissemination of valuable

production, management, and cost information among citrus

growers. It has helped create a dynamic source of information

that would ultimately lead to improved production and

management decision making.

A citrus production cost information system (COINS) with

three components, namely, a cost entry program, a summary and

comparison program, and a decision support system for

extension and management recommendations, has been designed,

developed, and evaluated. The system collects, averages, and

summarizes citrus production costs. It provides a means for

growers to compare their operations with industry averages.








An herbicide and weed guide gives appropriate management

recommendations and associated costs. Citrus growers from

several areas in Florida utilized the system and evaluated its

operational performance and its qualifications as a decision

and management tool. The system's components were found to

operate properly, and the system as a whole was qualified to

be a useful management tool.

The data entry component successfully collected,

organized, and stored production cost information for numerous

grove operations. The summary and comparison component

successfully averaged production costs for specific areas and

types of operations. It allowed for the comparison of

individual grove costs with industry averages. The herbicide

and weed guide integrated cost information with extension and

management recommendations through an expert system. The

successful integration of production costs with management

recommendations demonstrated the system's success as an

effective decision and management tool.

The utilization of a database management environment

provided a comprehensive tool for the development of the

information system. The techniques and design methods used in

development and integration of various components produced a

highly dynamic and flexible system.














CHAPTER I
INTRODUCTION

Justification

Citrus growers in Florida are continually striving for

improved methods of production in order to increase yield and

minimize costs. They are challenged by factors such as

weather conditions, consumer demand, market prices, and

foreign competition. They must overcome these obstacles by

using the best production and management practices available.

The cost of producing citrus is partly accrued through

the various grove operations performed during each season.

Production cost records are maintained by most operations as

a means of keeping track of expenditures, and for record

keeping and tax purposes. The amount and detail of cost

information recorded depends largely on the size of the

operation, and on the importance placed on such records by the

operation manager. In some instances grove cost records are

hand written, lack any detail, and are mainly used for year-

end tax purposes. More detailed costs are kept by some

growers who use computers to maintain data on a daily basis.

In some instances, growers owning small operations delegate

the record keeping task to caretaker operations and

cooperatives who usually have the means to maintain large

amounts of data on computer.









2
Growers seldom have the opportunity, however, to compare

their production costs with industry averages in a way that

will help identify problem areas. Database systems currently

being used by large operations to help maintain cost

information rarely offer any special data analysis or

comparison features. Most specialty financial software

available for the citrus industry deals strictly with farm

financial records. In most cases the software is used for

record keeping purposes. In some cases financial analysis of

information is performed by the software, and results in an

overview of the operation's financial status, but does not

deal with operational costs of individual grove practices.

Information sharing among citrus growers would enhance

the dissemination of valuable knowledge on production costs

and management practices. A large database system, Florida

Agricultural Information Retrieval Systems (FAIRS) (Johnson

and Beck, 1986), has been developed to provide growers,

researchers, and extension specialists valuable information on

management and extension practices, as well as technical

information on most topics related to citrus. The information

in the database is in the form of text and graphic screens,

and is updated with current data on a regular basis. Other

information sources such as technical publications, extension

specialists, and researchers, also provide growers with a

consistent source of information. According to Smith et al.

(1988) sources of new technology and production methods









3
include farm magazines, discussion with human expert or

neighbor, and extension service.

There is, however, a lack of availability of adequate

information associated with production costs for citrus.

There is also a need for a means to communicate this

information among growers. Knowledge sharing between citrus

growers is the key to achieving a dynamic database of

information. Using the "dynamic" approach would result in a

more specific and more usable database that would result from

simply keeping publications related to a specialist's subject

matter area on-line (Jones and Hoelscher, 1987). Furthermore,

no attempts have been made to integrate citrus production

costs with extension and management recommendations. Such an

integration would enhance growers' ability to make better

management decisions.

There is no reference in the literature to a dynamic

agricultural information system where production costs are

collected, pooled, summarized, and linked to appropriate

extension and management recommendations.



Objectives
The general objective is to provide citrus growers in

Florida with a means of improving their production practices

through the enhancement of knowledge and management decision

making, by providing an economic basis for extension and

management recommendations. This is achieved by developing









4
and integrating a computer based citrus production cost

information system with extension and management

recommendations. Information in the system develops and

expands through the addition of new management and extension

information as well as annually contributed production cost

information.

The specific objectives are as follows:

1) to identify and classify grove operations associated

with citrus production in Florida;

2) to design and develop a dynamic database system for

collecting and storing production cost information for Florida

citrus operations;

3) to design and develop a system to query the cost

database, average and summarize costs, and generate per acre

cost and returns reports;

4) to provide a means for comparing individual grove

costs with industry averages to aid in problem area

identification.

5) to design and develop a prototype decision support

system to integrate cost information with extension and

management recommendations.

6) to verify the information system's capabilities to

enhance the communication of knowledge between citrus growers

through the accessibility of citrus production cost

information and related management and extension

recommendations.









5
The remainder of this dissertation is organized as five

more chapters. The review of literature in Chapter II gives

an overview of management information systems and decision

support systems. It also deals with expert systems and cost

systems as related to agriculture.

Chapter III discusses the design requirements of the

system, including preliminary research, hardware, software,

and functional requirements.

The design and development of the information system

consisting of a database system for handling citrus production

costs and a data summary and comparison are discussed in

Chapter IV.

Chapter V deals with the design and development aspects

of the prototype decision support system for herbicide

application. This chapter also describes the means by which

production cost information is integrated with management

recommendations.

Chapter VI reports the results and procedures of the

evaluation and testing of the coins system. The system was

evaluated by various citrus growers, researchers, and citrus

extension specialists.

Finally, Chapter VII presents conclusions and

recommendations for future work in the area of agricultural

information systems.














CHAPTER II
REVIEW OF LITERATURE


Management Information Systems

Drechsler and Bateson (1986, p. 53) define a management

information system as "an information system that provides a

manager with information on the activities and pertinent

interrelations about the current status of the

production/operation system over which he has control". Smith

et al. (1985, p. 2) define management information systems as

"integrated computer based systems which provide information

to support the operational and decision making functions of

management". Murphy (1989) claims that in today's terms, MIS

(Management Information Systems) are simply databases. This,

he says, is an appropriate title since they are storehouses of

all information about a specific subject.

Information systems have become an important decision aid

and management tool for operation managers. They have become

an essential part of many organizations (Smith et al., 1985).

Both decision support systems and expert systems should be

included as subsystems of an overall management information

system concept.









7
Agricultural Information Systems

Today's businesses design and develop powerful

information systems for themselves. The farm community

deserves comparable quality power. To achieve this goal,

farmers must become involved in the design and development of

their systems (Murphy, 1989).

Computer-based systems that are designed specifically for

agriculture must meet the farmers' challenge (Stone et al.,

1986). Agricultural information systems may include a

database representing current knowledge about crop systems,

and may have the capability to access historical data records

and the ability to analyze data and expand on the analysis

presented to a user when requested. An information retrieval

system designed for farm use would encompass a collection of

resources including record keeping facilities such as

accounting information and expert systems (Beck, 1988).

The Jackson County Cooperative Extension Service (CES)

office, for example, uses an information management system to

collect, store, and retrieve extension information, and

provide it to clients in the form most suitable for their

needs (Heatley, 1986). The system saves time and effort that

would have otherwise been used to manually order and store

information, and pass out CES publications. The information

management system uses a database management system to

retrieve blocks of text.








8
Agricultural Technoloyv Transfer

Technology transfer from researchers and other

information sources to producers remains inadequate. The main

objective of agricultural research is to improve the

efficiency of farmers by providing them with valuable

information which they need to make cost effective production

decisions (Smith et al., 1985). The technology of information

communication is made possible through management information

systems. Murphy (1989) predicts that in the next decade or

so, companies will put a lot of effort into 'networks' that

link their information sources together. This will ultimately

enhance the information communication process.

The process of knowledge transfer from researcher to

producer can also be achieved with the construction of

interactive computer based decision support systems (Smith et

al., 1985). Such systems use knowledge to create valuable

information. In order for decision support systems to be an

effective communication tool, various components such as

simulation, information analysis, and problem solving models

must be integrated within a single framework which can be

effectively accessed by different user levels. Lal et al.

(1987) state that an effective means of transmitting knowledge

from technology generator to technology user would be a

process that would permit the end user to question and seek

clarification on the recommendations given.










Agricultural Cost Systems

The basic purpose of cost systems is to generate cost

information. Cost information may be used to satisfy certain

managerial requirements. One objective of cost systems is to

assist management in controlling costs (Shah, 1981).

Management is in constant need of timely and reliable

information on costs incurred by various responsibility

factors. Such information will provide management with

decision making support, and aid in the process of cost

reduction. Shah continues to say that a major concern of

management is reducing costs while maximizing profits. A cost

reduction process requires knowledge of significant cost

items, the cost of major activities, the identification of

controllable and non-controllable costs, and the effect of

cost reduction in each activity on revenues and profits.

In agriculture, growers must know their production costs

in order to be competitive and profitable (Tripeppi and

Kucher, 1988). Many microcomputer programs have been

developed to provide growers with various types of management

information. In some cases, analysis of information is

provided in order to give management further insight into the

financial position of their operations.

Olson et al. (1986) discusses an annual report that

summarizes individual farm records for farms in South Eastern

Minnesota. The report shows averages as well as high and low

ranges. Whole-farm information as well as enterprise costs









10

and returns are reported. At year end, individual farmers can

compare their operation to the information provided in the

report to find areas that need management attention and areas

which have above average performance. Some computer aided

farm business analyses is performed annually using IBM FINANX

software by the extension service, and a summary of individual

analyses is prepared.

A microcomputer program has been developed to calculate

production costs based on various inputs involved in growing

crops (Tripepi and Kucher, 1988). The user enters costs for

land, buildings, equipment, general overhead expenses, and

cultural practices. The program then produces cost summaries

for capital requirements, annual fixed costs, variable cost

per hour of equipment, estimated costs of production, and

final price per plant. Using this information, growers can

compare their prices with those of their competitors and can

therefore determine crop profitability. The program saves

hours of manual collating and adding production costs and

eliminates math errors due to human mistakes.

Another computer program, PPCAM was developed to manage

plant production data and enable the user to predict

production and profits (Power et al., 1989). Using input

data, PPCAM generates labor, material, energy, and indirect

cost information. The program can also generate, store, and

retrieve reports, summaries and graphics of various

parameters. PPCAM uses an integrated project-file language









11

using several spreadsheet templates and databases to create a

user friendly environment.

ABC Systems (1989) has created a comprehensive computer

program for agricultural record keeping. The program a mainly

an accounting software package that handles information for

budgets and plans, cash flows, cash forecasts, cost of

production, and offers results and balance sheet analysis.

An economics oriented program is discussed by McGilliard

and Clay (1986). The microcomputer program for decision

analysis (DECAL) was written to provide a relatively easy and

flexible method of analyzing investment decisions. For each

decision, DECAL produces a one page report showing business

measurements of net cash, present, future, period, and annual

values, as well as benefit/cost ratios, return on investment,

and payback period.


Decision Support Systems

Decision support systems are an emerging area of

research. They can combine database management systems and

the branch of artificial intelligence known as knowledge

representation. (Beck, 1988). According to Barrett and Beerel

(1988), conventional computing is concerned with handling

information which might subsequently be used in the decision

making process. Data processing organizes data and transforms

it from one form into another. Here, information is still at

a very basic level. Decision support builds on this by









12

allowing the manager to view data at a level which is

convenient to him.

Producers can use decision support systems to analyze and

help solve agricultural problems (Smith et al., 1985). A

decision support system should be able to provide appropriate

information to decision makers in order for them to operate

from a wider knowledge base than they do at present. A

decision support system discussed by Pruss (1989), utilizes

crop production information in the form of raw data. The

system collects, organizes, and summarizes information which

is later evaluated and interpreted into knowledge used for

crop management decisions.

Decision support systems may include expert systems.

Expert systems help alleviate some of the difficulties of

using decision support systems with a non-expert client.

Expert systems can help users select the most important

information for certain questions (Love, 1988). Expert system

technology can be applied at different points in the decision

support system-user interaction process, data collection,

identifying the problem, interfacing with the decision support

system, guiding the user through decision support system

output, assisting the user in "what if" applications of the

decision support system, helping the user select pertinent

output and decision weights, and analyze output. It is

important to ascertain at what point this application should

occur.











Expert Systems

Holt (1988) defines an expert system as a computer

program that enables a computer to mimic the logic of an

expert in diagnosing problems, selecting alternatives, giving

recommendations, and managing operational systems. Expert

systems are rule based and reason from one rule to the next,

gathering information until the system is able to recommend a

decision or provide advice (Helms et al., 1987). Expert

systems are highly interactive and generally easy to use (Lal

et al., 1987). An expert system user can ask the system

questions, change assumptions and even ask for the reasoning

behind answers given.

The biggest barrier to agricultural productivity is the

knowledge gap that lies between researchers and growers. An

expert system approach is ultimately an excellent way to

remove this obstacle (Rudd et al., 1986). Expert systems may

be used to enhance the capabilities of the researchers and

others responsible for the technology transfer process (Lal et

al., 1987).

Expert systems may be used as tools for summarizing

information and knowledge, diagnosing problems, and for

identifying specific objects and conditions such as weeds and

diseases. Expert systems can also be used as a teaching tool

for non-expert users (Holt, 1988).

McKinion and Lemmon (1985) discuss the role of expert

system technology in agriculture. They claim that the first








14

opportunity for using expert system technology in agriculture

is with integrated crop management. Expert systems would take

the form of integrated crop management decision aids which

would encompass such disciplines as irrigation, nutritional

problems, fertilization, weed control, cultivation, herbicide

application, insect control and insecticide and/or nematicide

application.

Expert systems may also be used for economic

applications. Expert systems designed to complement farm

financial records and planning systems hold considerable

promise as decision aids (Love, 1988). Producers can use

expert system technology to assist in synthesizing information

to decide the financial state and performance of their

operations.



Expert System Applications

System development in agricultural and natural resource

management applications has mirrored the growth in recent

years of the development and use of expert systems in product

design, resource management, and logistics (Lambert and Wood,

1988). A survey of agricultural expert systems currently

under development or available for use was completed. The six

major subject areas surveyed were crop and livestock

production, financial analysis, general shells, marketing,

natural resource management,and other areas. Most of the

applications surveyed are diagnostic in nature. However, a








15

few are dedicated to advising users on a variety of concerns

from irrigation scheduling to grain marketing to enterprise

selection. The survey reveals five programs in the area of

financial analysis.

A Financial Analysis Review System (FinARS) (Boggess et

al., 1989) is written in INSIGHT 2+ (Information Builders,

Inc.). The system provides an evaluation of the financial

health of a farm business. It is designed to provide an

initial assessment of the overall financial health of the

business.

In addition to its capabilities as a diagnostic tool for

farmers to provide initial interpretation of their farm's

financial situation, FinARS can also be used as a tool for

teaching financial analysis concepts to students, county

agents, lenders, and farmers.

Lambert and Wood's (1988) survey also included four other

expert system applications related to economics that are worth

mentioning. A Budget Planner written in PASCAL is an

enterprise budget generator that includes a full enterprise

analysis. A farm financial document analyzer written in LEVEL

5 is designed to assist producers with financial management

decisions for Michigan dairy farms. An agricultural financial

analysis expert system is available to give information on a

farm's current year performance, financial condition, and debt

repayment ability. And finally, a farm loan advisor was

written to evaluate farm loan applications.









16

Helms et al. (1987) discuss a farm level expert system

that provides advice regarding farm program participation and

changes in farm policy variable. The description and

application of the Farm Policy Advisor or FPA was demonstrated

on a Southern Blacklands hypothetical farm in Texas.

Richardson et al. (1989) describe an application whereby

mini-expert systems are called to develop data from user

inputs to be used in a database. A main program CARMS

(computer assisted records management system) allows users to

build data sets needed by CIRMAN (crop insurance risk

management analyzer). The expert system CARMS leads the user

through a series of steps to develop a large sophisticated

database required for a simulation model.

Linker et al. (1990) discuss a herbicide recommendation

program that allows the weed management expert to build a list

of recommended herbicides or herbicide mixtures based on weed

species and soil type. The program, written in C, allows the

user to obtain a herbicide recommendation from a particular

weed problem.

Batchelor et al. (1989) describe a prototype expert

system developed to aid in soybean insect pest management

decision making. Crop status and insect population

information are provided by the user. The decision support

system, SMARTSOY, runs a crop growth model to determine

subsequent damage to the crop. The crop growth model,

SOYGROW, predicts growth and development and final yield of









17

soybean based on daily weather data for specific soils (Jones

and Hoelscher, 1987). Once SOYGROW is run, SMARTSOY gives

cost effectiveness of insecticide applications as well as rate

and type of insecticide are given.



Database Manaaement Systems

Lal et al. (1987) state that the utilization of database

management has made a definite contributions to the technology

process. Database management typically involves the mechanics

of storing and retrieving large amounts of data (Beck et al.,

1987). Efficient database design is essential for file

organization, indexing, rapid transaction and query processing

rates, and allowing multi-user access and sharing of data.

Database management also addresses problems such as data

security, maintenance of data accuracy and integrity, and

creating data storage separate from the application programs

which use the data.



Database Design

Proper database design is imperative when creating an

efficient data management environment. Brathwaite (1985)

discusses some essential reasons behind careful database

design including data redundancy, application performance,

data security, and ease of programming.

Hierarchical Models. Hierarchical data models are based

on tree-like structures made up of nodes and branches, where









18

the highest node is called the root, and succeeding lower

nodes are called children (Brathwaite, 1989). In hierarchical

models, trees are constructed using a father-son approach

(Chorafas, 1989). Hierarchical models can range from fairly

simple such as in the case of one-to-one or one-to-many

relationships, to the more complex many-to-many relationships.

Brathwaite (1989) and Chorafas (1989) both discuss some

advantages and disadvantages of hierarchical models. A major

advantage is the existence of proven database management

systems that utilize the hierarchical model. Another

advantage is the simplicity and ease of use of hierarchical

models which ultimately will facilitate their employment and

utilization by data processing users. Other benefits of

hierarchical models include their ability to reduce data

dependency, and their capability to efficiently represent

decision support system data (Hopple, 1988).

Both authors agree that one of the major disadvantages of

using hierarchical data models as a basic structure is their

lack of flexibility. This presents difficulties during

insertion and deletion operations. Due to strict hierarchical

ordering insertion and deletion of files or entities may

disrupt the tree structure. Also, deletion of the a parent or

father node will result in the deletion of the children

associated with that node. Accessibility of information is

also mentioned as a disadvantage since a child node is

accessible only through its parent node.








19

Relational Models. Relational database models store

information in tables of rows and columns (Singh, 1985). The

association of rows and columns in a table is the

characteristic that gives the relational data model its name

(McNichols and Rushinek, 1988). McNichols and Rushinek

discuss that one advantage of the relational model is that

operations on tables can be defined mathematically, allowing

precise and unambiguous data retrieval. Maier (1983) states

that another advantages of this type of data storage is its

uniformity.


Data Security

An important part of database design is the incorporation

of a system to maintain data security. In an information

system with multiple users, a method of ensuring that only

authorized persons can access certain data is essential. Most

systems use a control of access method whereby a user's

identity is verified, and according to his priority, the

system assigns him access to certain information (Delobel and

Adiba, 1985). Any users attempting to enter the system must

identify themselves and then authenticate the identification

(Brathwaite, 1985).

Harington (1988) states that the first step in on-line

database security is to identify the user. However, in single

microcomputer operation systems there are generally no

individual user accounts or work areas. Therefore, the









20

computer assumes that everyone who has access to the keyboard

is authorized to use the machine. Consequently, security must

be handled at a lower level by the database management system

itself.

DBASE III Plus (Ashton-Tate, Inc.) has no features

specifically designed for data security, but it is possible to

impose some measures of safety through an application program.

However, even with the use of a password system through the

development of an algorithm, there is nothing that will

prevent a knowledgeable user from running DBASE III Plus and

accessing data by using ordinary program commands.

In many businesses, however, DBASE III Plus users have

not been trained to work directly with the database management

system environment. The majority are not computer

professionals, but people trained in other areas who use

computers to help do their jobs. These users work with

application programs whereby interaction with the system is

made through menu-driven interfaces and system screen forms.

In this case, users can be required to supply a password which

governs their access to the program.



Integration of Knowledge Systems and Databases

The integration of expert systems with other conventional

software would be beneficial (Holsapple et al., 1987). Some

expert system environments have a limited ability to import

data from external software. This ability does not, however,









21

replace integration, which provides a comprehensive

environment that makes all business computing abilities

available for use at any time, individually or in tandem.

Expert systems should be viewed as a supplement rather

than a replacement for existing computer technologies like

database management systems (Lal et al., 1987). Stone et al.

(1986) describe a Farm Level Expert System (FLEX). FLEX is an

integrated system made up of expert systems linked through a

common global memory and a database of interrelationships.

The integration process involves several stages (Harmon and

Sawyer, 1989).

Harmon and Sawyer state that first, it is important to

determine whether or not a database is necessary. Databases

may be used for maintenance purposes whereby they are used to

externalize the parts of a decision support system that change

frequently. Databases can also be used as tools for

information storage, retrieval, searching, and querying. A

database management system is often better suited for

efficient search of large volumes of structured information

than a expert system is. Data sharing with other

applications, and other users, as well as security

maintenance, are other reasons that necessitate the use of a

databases in decision support system.

Second, Harmon and Sawyer say that the role of databases

in decision support systems should be evaluated. An expert

system can be used in two different ways: as a front end for









22

a database, asking questions and then initializing database

queries, or as a back end for a database, taking the results

of the query and analyzing them with rules in order to make

recommendations. Jones and Hoelscher (1987) describe a method

of integration whereby an expert system provides an initial

information sorting process, and then accesses a main database

for more detailed information.

When an expert system is used as a front end for a

database, the consultation proceeds in the normal way, with

the system asking the user questions and using rules to reach

a final conclusion. Once the system reaches a conclusion, it

takes the extra step by constructing a database query and

searching the database for an even more specific

recommendation. The database can hold information that will

further elaborate on the recommendation. Gallagher (1988)

discusses that the processing of data by decision support

systems may be very inefficient if they do not have the

ability to work with information proficiently. A solution to

this problem requires the system to request processing of data

by other programs that are specifically designed to accomplish

that task. That is, ask the data base management system to

search databases and summarize data.

When used as a back end to a database, an expert system

takes the database output and uses the information as input.

It then makes judgements based on it, and presents specific

suggestions to the end user. Back end applications are









23
usually used to take the raw data drawn from a database query,

process it further, and convert it into specific

recommendations. This approach is very effective for

providing support to individuals who have difficulty

interpreting the database query output. Expert systems are

used to build on the results of data processing and decision

support by assisting the user with the interpretation of data

to formulate responses (Barrett and Beerel, 1988).



Summary

The citrus cost information system discussed in this

dissertation integrates several of the technologies addressed

in this chapter. A database management environment is used in

conjunction with an expert system to provide production cost

information and extension and management recommendations.

The cost information system is unique in its ability to

provide industry averages in a way that is useful as a

decision making tool. The expert system provides information

on not only extension and management recommendations, but on

the costs associated with these recommendations.














CHAPTER III
SYSTEM DESIGN CONSIDERATIONS


During the implementation and use of the Florida

Agricultural Information Retrieval System (Johnson and Beck,

1986), Florida citrus growers expressed a need for information

on production costs associated with citrus. They needed an

economic basis for extension and management recommendations.

An information system was needed to provide growers with the

means to contribute production cost information for their

operations, and in return receive summaries of industry

averages. The system would not only provide cost summaries,

but would also integrate those costs with appropriate

extension and management recommendations to provide a

comprehensive decision making tool.



Preliminary Research

Existing Information Sources

The first phase in this project was to determine what was

already available for the citrus grower in the form of

production cost information. It was found that an annual cost

survey has been carried out since 1973 by Ron Muraro, a

researcher at CREC (Citrus Research and Education Center) in

Lake Alfred, Florida (Muraro and Matthews, 1988). Muraro's








surveys are designed to collect custom charge rate information

from citrus caretakers in different citrus producing areas of

the state. Written surveys are mailed out once a year to

participating caretaker operations. When all information is

collected, the relevant data is organized and summarized in

annual publications. Although other agricultural cost surveys

have been carried out in the past in areas such as potato

production, and woody nursery businesses, Muraro's citrus cost

surveys are the only ones still being carried out on a regular

basis.



Existing Agricultural Software

The next step in the preliminary stages of this project

was to determine what types of software packages were being

written for Florida citrus operations. Several programs were

available through the IFAS (Institute of Food and Agricultural

Sciences) software distribution office. After reviewing what

the IFAS catalog had to offer it was determined that most of

the programs were not written specifically for the citrus

grower. Some of the programs required the user to have a

considerable amount of computer experience, and it was often

necessary for the user to have a certain software package or

language such as LOTUS 123 or BASIC to run the programs.

Based on visits to software exhibits, agricultural trade

shows, and discussions with citrus experts, commercial

software proved to be mainly designed for payroll and record








26

keeping purposes and did not offer the grower any decision

support capabilities.



Design Reauirements

The User Interface

One of the chief design requirements was a user-friendly

yet detailed environment for the user to work in. The

development of a user-friendly interface is an important part

of any system development (Andriole, 1986). Andriole states

that the user interface is the single most important part of

today's microcomputer systems. Its design must be approached

carefully, and should not be left to chance.

Peart (1988) suggests specific guidelines for the design

of user friendly programs. Menu selections should be used to

obtain necessary words or strings from the user, rather than

requiring typing. The user should be given new information

and help when needed. Previously entered information should

be kept on the screen as long as possible. A user should

always have a way reexamine his data and go back and change

it. And finally, files should be designed to allow the user

to refer back to what he did in his last session.

Lal et al. (1990) discuss an interface program designed

to work in conjunction with an expert system. Specially

designed screens use pop-up menus with user defined options

for data selection. The program acts as an information

manager for an integrated decisions support system, FARMSYS.








27

Once data items are defined, they need not be entered again.

This ensures consistency of data and facilitating rule

handling, error-checking, logical decisions, and searches

during program operation.

The system user was assumed to be a novice computer user.

The program had to be easy to use, self explanatory, flexible,

and expandable. Paller and Laska (1990) stress the importance

of ensuring that a system maintains enough flexibility and

responsiveness to meet the real and changing needs of an

organization.



System Hardware and Software Requirements

It was felt that the majority of citrus growers in

Florida that utilized microcomputers in their operations owned

IBM or compatible machines. The system needed to run on the

hardware already available in order to avoid additional

requirements.

Software specifications required a stand-alone system.

As discussed above, a large number of the programs already

available through IFAS required additional software to run.

This often necessitates costly software packages to be

purchased by the user. If such supporting programs were to be

distributed with the information system, then the need for

expensive licenses would arise.










Functional Requirements

In order for any computer based system to be successful,

it must fulfill two requirements. It needs to perform the

tasks it was originally designed for, and it must offer the

end user enough benefits to justify the time invested in its

use. Any system must provide the user with a new or improved

method of performing a task by replacing outdated methods or

technologies and enhancing productivity. Software systems

must also provide information that complements or supplements

information already available to the user through conventional

means.

A citrus cost information system would require large

amounts of data to be contributed by the user. It is often a

labor intensive and time consuming task for a grower to

contribute production costs for several of his groves by

entering detailed information into a computer. In the case of

this project, the system needed to provide the user with an

incentive to encourage its utilization. This is achieved by

providing the user with valuable information in the form of

industry averages and management recommendations, and

integrating the two in a unique and beneficial manner. T h e

system would also offer the user the chance to compare his

operational costs with the industry averages in order to

identify possible problem areas.














CHAPTER IV
COINS CITRUS COST INFORMATION SYSTEM


An Overview of COINS

A Citrus Cost Information System (COINS) was developed to

collect, manage, and report costs associated with citrus

production in Florida. The program consists of three main

sections. The first is a data gathering program for

production costs for various areas of the operation. The

second section is a summary and comparison program which

provides detailed summaries of averaged production costs, and

allows the user to compare production costs for one of his

groves with industry averages. The third section integrates

production cost information with extension and management

recommendations. The system for integrating management and

cost information (SIMON) is discussed in more detail in

Chapter V.

The following section discusses COINS in more detail,

including software and computer languages used for

development, the overall program structure, and program

execution.








30
DBXL as a Database Management and a Programming Environment

To write the COINS programs, software which combined

database management with a comprehensive programming

environment was needed. Two relational database packages were

considered, DBASE III+, and DBXL (WordTech Systems, Inc.) a

similar database program. Both packages offered a database

management environment, as well as a programming language with

which to develop applications.

DBXL was chosen over DBASE III+ for its increased

flexibility and lower cost. DBXL offers an extended language,

and an added windowing feature which allows multiple windows

to be generated. This feature was found to be helpful in

designing menus, pop-up help screens, error message windows,

and in improving the overall quality of the user interface.

QuickSilver (WordTech Systems, Inc.) was used to compile the

database applications to create fast stand-alone programs.

This allowed COINS to run independently of dBXL. Thus

circulation of the programs would not infringe on any

copyright or distribution laws.



Program Development Tools

COINS was developed on a Compaq 386 computer running

under an MS-DOS environment. A memory resident editor was

used to write the programs. Windowing environments such as

Microsoft Windows 386, and Quarterdeck's Desqview 386 were

used to load and run all the necessary software








31

simultaneously. This allowed the programs to be written with

the editor, tested in a dBXL environment, compiled with

QuickSilver, and tested, with great efficiency.



Program Requirements

The system was designed to run on any IBM-PC, XT, AT and

compatibles as well as 386 based machines. Hardware

requirements include a minimum of 640 kilobytes of free RAM,

a hard disk drive, and a color monitor. The COINS program

files occupy approximately one megabyte of disk space. The

data files require 60 kilobytes of disk space per grove. An

additional 50 kilobytes of free disk space per grove is

recommended to accommodate the temporary files during

execution of the data summary and comparison program.

Program execution speed depends on the hard disk drive

access times, the microprocessor speed of the computer being

used, and on the presence or absence of a coprocessor chip.

Execution is fairly slow on PC and XT machine, and fastest on

386 computers.



Program Structure

COINS incorporates numerous database files and three main

programs, namely a data entry program, a summary and

comparison program, and a decision support system. All

programs are accessed through a main menu program written in

Turbo PASCAL (Borland). When COINS is started, the main menu








32

is loaded into memory and remains in memory until the session

ends. When execution of a program is complete, control

returns to the main menu (Figure 4-1).



Password System

COINS utilizes a password system that serves three

purposes. The first is to protect data files against

unauthorized access. Grove cost information can only be

entered or retrieved when the proper password is used. The

second purpose is to insure that a user has entered cost

information for at least two groves before he is allowed

access to the data summary and comparison program. And

finally, passwords along with grove ID numbers are used as a

means of identifying groves in the database.



The COINS Database

Database Structure

The first step in developing a database to hold citrus

production costs, was to establish a list of grove practices.

Such a list was already available in Muraro's surveys. Muraro

listed ten major grove practices (cost categories), each with

its own list of sub-categories. Using Muraro's list as a

basis, a modified set of cost categories and sub-categories

was created. Each category had five different types of costs

associated with it; overall, labor, machinery, management,

and material costs. Database files used for cost information


















































Figure 4-1


COINS Program Structure








34
storage were designed for maximum flexibility. They allow the

user to enter data at various levels of detail based on the

amount of cost information available. This is achieved by

arranging production cost categories in a hierarchical form.

There are ten main cost categories, namely cultivation,

dusting, spraying, frost protection, young tree care,

irrigation, removing trees, fertilizing, pruning, and other

operations. Each category has one or more levels of sub-

categories associated with it (Figure 4-2).

The ten main cost categories appear on the first screen

of the data entry or summary programs. This level of

categories in the hierarchy is referred to by the database as

level 1. Each level 1 main cost category has at least one

level of sub-categories. These subcategories are referred to

as level 2 and level 3. If more detailed information is

available on a particular grove practice, the user may expand

to a sub-category level and enter or view cost information.



Data Storage

Information on grove characteristics and grove costs are

stored in twelve separate data files. One file is specific to

grove characteristics, and is used for verifying passwords, ID

numbers, and to reference information. Production cost

information is stored in eleven main data files; one level 1

file, and ten level 2 files. dBXL data file structures are

shown in Appendix A.











- CULTIVATION-


--Hand Hoe/Hand Labor
--Machine Hoe
- Rotovate


-- 7"


-Disc--
-- Mow
--Herbicide -Of
--Temick/Nemacur --5
-- Plow --9-
--Backhoe --1
-Water Furrow Disc --V
-Water Furrow Cleaner --I
-Vine Puller --S
--Miscellaneous


- DUSTING Ground Application
-Aerial Application


- SPRAYING i Ground Application
-Aerial Application----


- FROST PROTECTION






- YOUNG TREE CARE-


- IRRIGATION-


General Costs
-Wind Machines
-Bank/Unbank Trees
-Tree Wraps
-Grove Heaters
-Other Methods


L--Spc
-St
--Tn
--Che


9-10"

fset/Side
-7'
-10'
5-16'
-Mower
n & Out
Lckle

)t Herb.
rip/Band
rnk-Trunk
im. Mow


--Dilute
--2X
-3X
- 4X
--6X
--10X
--15X
-Hand Spr.
-Boom Spr.
--Other


Reset Care
Planting Trees --Fixed Wing
Irrigation -Helicopter
Ring Trees Other
Fertilizing---
---Hand Spreader
-Fert. Spreader


- General Costs Perm. Overhead
- Repair & Maintenance Traveling Vol.
- Inject Chemicals -Stationary Vol.
- Water Pump --Perfor. Pipe
- Water Slinger --Microsprinkler
- Water Truck --Drip System
--Water Trencher
- Rotary Ditcher/Auger Fertil.
--3-Wheeler for Micro. --Herbicide
--Miscellaneous --Pesticide
Fungicide

Hand Labor
Mechanical


Production Cost Hierarchy


Figure 4-2
















- REMOVING TREES Tractor
-- Bulldozer
---Front End Loader
Miscellaneous

- FERTILIZING Liquid Boom Appl. Nitrogen
Dry (Bulk) -Mix-Fertilizer
Dry (Bag) -Mix Fertilizer
Liquid Nitrogen & Herbicide
--Lime and Dolomite
-Miscellaneous

- PRUNING General Pruning
Hedging Single Side
Topping -Tractor
Removing Brush- Self Prop.
Miscellaneous
--Double Sided
- OTHER OPERATIONS Tractor
-_ Self Prop.
--General Repairs
__--M G. Cvf% T- r ac tr .-


--Energy Costs
-Ground Bypass
-Truck with Driver
-Tractor with Driver
-Power Saw
-Front End Loader
--Push Brush
--Bush Hog
--Rotary Ditcher/Auger
- Mound Builder
-- Middle Buster
--Skilled Labor
--Mechanic Labor
--Supervision Costs
-- Contracted Services
--Accounting Services
--Miscellaneous


-I Self Prop.


_-_Haul Out Of Grove
Chop Brush


Figure 4-2--Continued








37
Supporting database files hold information on fruit

varieties, grove locations, number of categories and sub-

categories, and sub-category locations within the database.

Other files include index files and query files. Each cost

information data file has a corresponding temporary file.

Temporary files are used for data entry and modification, and

for data storage after summaries are generated. All user

interaction with the database is done through the temporary

files. This allows for quicker access, and protects the

integrity of the main data files.

When the data entry program is run, appropriate

information from the main data files is copied into the

temporary files. Hence, if information for a new grove is

being entered, information for a grove with costs set to zero

(default grove) is copied from each main data file into its

corresponding temporary file. Upon completion of data entry

and data modification, cost information is appended to the

main data files from the temporary files. On the other hand,

if information from an existing grove is needed, corresponding

cost information for that grove is copied from the main data

files into the temporary files, and displayed for editing.

When editing is complete, the new information on the grove

replaces the old data in the main data files.

Temporary data files are also used when the data summary

program is run. Cost information from the main data files for








38

all groves to be included in the summary is averaged, and the

results placed in the temporary files for viewing.



Data Entry Proaram

Program Design

The data entry program was designed to look and function

like a spreadsheet (Figure 4-3). A cursor bar is used to

select between ten production cost categories. A menu bar is

used to select among ten available functions. Functions are

used to access the different features of the program. To

select a function, the user moves the cursor bar and presses

the enter key. Alternatively, a function may be selected by

pressing a specific highlighted character found in each

function name. An explanation line below the menu options

helps to identify what each function does. The following is

an explanation of the ten functions.



Program Functions

Edit. Data is entered using the edit function. When

edit is selected, the production category on the highlight bar

is highlighted in red, and a help window appears at the bottom

of the screen. Costs for labor, machinery, management, and

materials may be entered for each category. When the editing

is concluded, values in the separate columns are summed, and

the results are placed in the overall cost columns. The

automatic summation may be suppressed by entering a value for























Variety:Hamlins,Valencias. Market:Fresh/Process. ID: 1


Grove Practice Overall Labor Mach. Mngmt Material
--------- Cost Cost Cost Cost Cost

Cultivation 0.00 0.00 0.00 0.00 0.00
Spraying 0.00 0.00 0.00 0.00 0.00
Dusting 0.00 0.00 0.00 0.00 0.00
Fertilizing 0.00 0.00 0.00 0.00 0.00
Irrigation 0.00 0.00 0.00 0.00 0.00
Removing Trees 0.00 0.00 0.00 0.00 0.00
Pruning 0.00 0.00 0.00 0.00 0.00
Young Tree Care 0.00 0.00 0.00 0.00 0.00
Frost Protection 0.00 0.00 0.00 0.00 0.00
Other Operations 0.00 0.00 0.00 0.00 0.00

New Data File

Edit Help Mode Expand Return Locate Update Comment
Product Back
Explanation Enter costs for this category.


Example Initial Data Entry Screen


Figure 4-3









40

the overall cost, or by using the mode function explained

below.

Helt. The data entry program features an on line help

utility. When the help function is selected, an options

window appears and allows the user to select among several

help topics (Figures 4-4a and 4-4b). Help topics include an

introduction to the purpose and uses of the program, a short

tutorial, an overview of the database structure, an

explanation of the ID number to identify groves, and lastly a

help utility that explains each function (Figure 4-5). The

user may also obtain help on functions by moving the menu bar

to the desired function, and pressing the Fl key.

Mode and Update. The mode function allows switching

between auto-calculate and manual modes for column summation.

When the program starts up, auto-calculate is on, and whenever

separate costs for labor, machinery, management, or materials

are entered, the program automatically adds these costs, and

places the total in the overall column. The user may select

manual recalculation in order to enter separate costs for the

labor, machinery, management, and materials, as well as a

separate overall cost. This feature is important when the

user reports partial costs for a category, and the overall

cost does not equal the total of the separate costs.

The update function is used to total the costs in the

separate columns when mode is set to manual. The user may
















































Main Help Menu


Help Menu


=> General Help Utility.

Help With Functions.

Quit Help.



Use the arrow keys to select a topic and
press . Press to quit.


Figure 4-4a


























Help Menu

General Help Menu


=z> What Does This Program Do ?
Getting Familiar With The Program.
Database Structure.
The Grove ID Number.
Quit The General Help Utility.



Use the arrow keys to select and press .
To return to the Main Help Menu press ESCAPE.


General Help Utility Menu


Figure 4-4b


























Help Menu

Available Functions


Edit Return Comment

Mode Locate Product

Expand Update Back


Press the appropriate letter for a function
to select it. Press to quit.


Figure 4-5 Function Help Menu









44
wish to total all costs, or just those in the current cost

category.

Expand and Return. Each of the ten main cost categories

is divided into one or more levels of sub-categories. The

expand function is used to access the sub-categories for more

detailed data entry. The return function allows the user to

exit from a sub-category screen and return to a higher level

in the hierarchy.

Locate. The locate function serves two purposes. The

first is to show a list of sub-categories available for a

particular cost category. This serves as a "map" for the

operation hierarchy (Figure 4-6). The second usage of the

locate function is an index for all grove practices available

in the database. This function is used to locate the main

category under which a particular grove practice is found.

For example, to find the main cost category for 'Energy and

Fuel Costs', the user searches through the index for the word

mowing. The index will indicate the main category as being

'Other Operations' (Figure 4-7).

Comment and Product. The comment function allows the

user to enter a comment for each cost category. Comments may

be used as notes or reminders. Similarly, the product feature

allows the user to leave a description of the product name, or

material name used in a particular grove practice. This

feature may be important if the program is expanded to include

























View Of Operation Hierarchy I


Level 2


Cultivation
Spraying
Dusting
Fertilizing
Irrigation
Removing Trees
Pruning
Young Tree Care
Frost Protection
Other Operations


Hand Hoe/Hand Labor
Machine Hoe
Rotovate
Disc
now
Herbicide
Temick/Nemacur
Plow
Backhoe
Water Furrow Disc
Water Furrow Cleaner
Vine Puller
Miscellaneous


Offset or Side
5-7'
9-10'
15-16'
V-Mower
In & Out
Sickle


Figure 4-6


Example for the Locate Function for Sub-
categories Under the Level 1 Cultivation
Category, and Level 2 Mow Sub-category.


Level 1


Level 3

























Cost Categories

Use the arrow keys => Energy and Fuel Costs
to browse the list. Fertilize
You may also search Fertilize Young Trees
by pressing the first Front End Loader
letter of a desired Frost Protection
category. Ground Bypass
Grove Heaters
Press to Hedging
return to the cost Herbicide
screens. Hoe
Inject Chemicals Irr.
Irrigate


Main Cost Category
--> Other Operations


Figure 4-7


Example for the Find Function for Energy and
Fuel Costs.









47
collection of names of pests, weeds, chemicals, or other

products used in the citrus operation.

Back. This function is used to return to the COINS main

menu. When the back function is used, all the main category

files are updated using the costs entered during the session,

and control is given to the main menu.


Entering Costs for a Grove

The following section describes the procedure for

starting and using the data entry program to input or edit

costs for a grove. Three options are available when the

program is started. The user may either define and enter

costs for a new grove, view or edit information for an

existing grove, or delete one of his groves from the database.

Each grove is assigned a grove ID number which is

utilized by the user and by COINS to identify the grove.

Along with the ID number, a grove can be identified by the

user's password. The user must have both the a password and

an ID number to enter or edit costs for a grove.

Editing an Existing Grove. If the user chooses to view

or edit information for an existing grove, the program

displays a menu of ID numbers corresponding to available

groves, and prompts for a selection (Figure 4-8). By moving

the menu selection bar to each ID number, the characteristics

of each grove are displayed. When a grove is chosen, the

program searches the database files, retrieves, and displays









48
the cost information found. It is important to note here,

that only those groves associated with the user's own password

can be accessed.

Defining a New Grove. To enter costs for a new grove,

the user must define the characteristics of the grove. These

include size in acres, fruit varieties, fruit market, yield in

boxes of fruit, grove location, and grove ID number. A grove

definition screen is used to enter the necessary data (Figure

4-9). A list of options is available from which the user

selects fruit varieties and grove location (Figures 4-10a and

4-10b). Choosing from a menu of options allows easy

selection, and ensures that no discrepancies occur.

Once the information on a grove is complete, the user is

prompted to enter a grove ID number. This number is unique to

the user's password. Two or more groves with the same ID

number may exist as long as they were defined by different

users.

Entering and Modifying Data. When the spreadsheet is

displayed, the user may enter or change data at any level in

the hierarchy. The arrow keys are used to move around the

spreadsheet, and a menu bar is used to select the functions

described in the previous section. When entry or modification

is complete, the data entry program updates all files, and

returns control to the main menu.

Deleting Grove Information. Often times a grove is no

longer active, or no longer belongs to the person who entered

























Florida Citrus Production Cost Survey Program


Grove Description Available
I Groves

Grove size (acres) ... 5.00
--> 40
Variety ... Hamlins, Valencias 240
530
Yield in boxes ....... 3745.00 10

Price per Box ($)..... 6.75

Grove location ....... Ridge

Grove ID number ...... 40


Figure 4-8


Example of Grove Selection Screen for Viewing
and Editing Costs


























Florida Citrus Production Cost Survey Program


Grove Description


Grove size (acres) ... 0.00

Variety ...

Yield in boxes ....... 0.00

Price per Box ($)..... 0.00

Grove location .......

Grove ID number ...... 0


Example Grove Definition Screen


Figure 4-9

























Florida Citrus Production Cost Survey Program


Grove Description Options

Hamlins
Grove size (acres) ... 5.00 Valencias
Navels
Variety ... e... Temples
Early/Mid
Yield in boxes ....... 0.00 Minneolas
Lees
Price per Box ($)..... 0.00 Murcotts
Novas
Grove location ....... Marsh
Robinsons
Grove ID number ...... 0 Duncan


Fruit Variety Selection Options


Figure 4-10a

























Florida Citrus Production Cost Survey Program


Grove Description Options

Ridge
Grove size (acres) ... 5.00 Interior
Ind. River
Variety ... Hamlins, Valencias Southwest
North
Yield in boxes ....... 2500.00 Central
Other
Price per Box ($)..... 7.63

Grove location ....... MEN.g

Grove ID number ...... 0


Location Selection Options


Figure 4-10b









53
its costs into the system. A grove may be sold, or in light

of the recent freezes, a grove may simply no longer be

productive. In such cases the user may wish to remove this

grove from the database. When the grove deletion option is

chosen, the user needs only to specify the grove ID number,

and the program will remove all costs and grove information

from the database.



Data Summary and Comparison Program

Program Design

Like the data entry program, the data summary program was

designed to look and function like a spreadsheet (Figure 4-

11). The main difference between the two programs is that the

data summary program does not support any editing features.

It is simply a tool for averaging costs, and displaying the

results. As in the data entry program, a cursor bar and a

menu bar are used to select categories and menu options

respectively.

The summary screen provides average per acre costs and

returns for the ten main cost categories, as well as both

levels of subcategories. Per acre average labor, machinery,

management, and material costs are also displayed.






















Averaged Annual Per Acre Your Grove #: 40
Costs and Returns


Cost Item Range of Costs Average Your Grove

Revenue 2786.84 5055.75

Cultivation 76.70 to 194.36 142.26 194.36
Spraying 138.20 to 183.00 167.36 180.88
Dusting 0.00 to 0.00 0.00 0.00
Fertilizing 69.47 to 241.22 173.12 241.22
Irrigation 5.14 to 261.65 169.86 261.65
Removing Trees 0.00 to 1947.98 978.46 1947.98
Pruning 0.00 to 472.70 472.70 0.00
Young Tree Care 71.14 to 270.98 138.48 71.14
Frost Protection 0.00 to 72.08 37.32 2.56
Other Operations 26.55 to 48.44 35.20 26.55
Total Per Acre Expenses 2279.56 2926.34
Per Acre Net Income 507.28 2129.41


Help Options Labor Machinery Management Material
Locate Expand Return Back
Explanation Access the help utility.


Example Cost Summary Screen


Figure 4-11










Program Functions

For the purpose of avoiding redundancy, only those

options that differ from the data entry program menu will be

discussed.

Help. The help option is similar to that in the data

entry program, but differs in the contents of the tutorial,

and explanation of functions.

Options. This function accesses a decision support

system discussed in Chapter IV of this dissertation.

Labor. Machinery. Management. Material. When these

options are used, the screen switches from displaying per acre

overall costs and returns to a screen showing only the

selected separate cost.



Cost Averaging

Li et al. (1990) discuss two major methods of retrieving

database information. One is a navigation-based method

whereby the user uses a series of menus to travel from one

area of the database to another. The second method is the

query based access to information. This method uses the users

questions to retrieve information. The COINS data summary and

comparison program uses the data query method to retrieve the

necessary data from the cost databases.

The program requires the user to specify parameters which

determine the types of groves that will be used in the

summary. Maximum and minimum grove sizes, fruit types, fruit









56
market (fresh or process), as well as location of the groves

must be specified (Figure 4-12). The query may be either very

general whereby only one or a few parameters are specified, or

very specific. In this way the user can tailor the cost

comparison to his own needs and interests. The program also

allows the user to specify a grove to compare to the averages.

If a grove is specified, per acre costs for that grove will be

displayed alongside the averages in the summary screen.

Once the query is defined, the program searches the

database for all groves fitting the query and averages the

costs found. All overall, labor, machinery, management, and

material costs for each category are averaged separately.

Data on highest and lowest costs in each category is

maintained and later displayed as a range of costs.

The program is designed to carry out a data summary only

if three or more groves fit the query. This protects the

confidentiality of the data, and prevents the user from

"extracting" costs for a single grove in the database by

specifying its exact characteristics.

Once the summary screen is displayed, the user may choose

to explore the average values in the different levels of the

hierarchy as well as view the individual labor, machinery,

management, and material per acre costs. A data analysis

option allows the user to compare his costs for an operation

to the averaged values, and invoke an expert system that aids
























Florida Citrus Cost Reporting Program


Grove Information Groves
Found

Minimum grove size ... 0.00 Found 3
Averages:
Maximum grove size ... 5.00 Size =
5.00 acres
Variety ... Any Variety Yield =
2314.67
Grove location ....... Any Region Price/
Box =
Grove ID number ...... 40 $ 6.02


Cost data is being prepared Wait


Figure 4-12


Example of Query Definition Screen During Data
Averaging









58
in the analysis and gives and management recommendations. The

analysis option will be discussed in more detail later.















CHAPTER V
SIMON SYSTEM FOR INTEGRATING
MANAGEMENT AND COST INFORMATION


A system was developed to integrate management and

extension recommendations with production cost information.

The System for Integrating Management and Cost Information

(SIMON) utilizes an expert system to access a database of

management information for a particular grove practice. Based

on each recommendation, the system determines an ideal cost

for a particular situation and compares that cost to actual

expenditures reported by the grower.

The SIMON concept can be applied to many area of the

citrus operation, or to other crop systems in addition to

citrus. But for the purpose of this project a prototype

system was developed to give management recommendations for

herbicide application and weed control on citrus. The user

defines a weed problem, by identifying categories and names of

weeds. These are chosen from a list of weeds commonly

associated with citrus. The program then searches a database

for appropriate spraying recommendations. The expert system

asks the user a series of questions to determine the proper

application rates and application times based on conditions

particular to his operation. After a herbicide recommendation

is given, the user has the option of determining the cost of

59









60
controlling the weed problem defined, and making a comparison

to costs in his grove. The system may be used in conjunction

with the COINS data summary program or as a stand-alone system

for citrus herbicide and weed control information.

The following section discusses the herbicide prototype

program and the SIMON concept in more detail. System

development, program execution, integration with the COINS

data summary program, and evaluation will be described.



Program Development

The benefits of an expert system based program for

management recommendations give it a clear advantage over

traditional methods of conveying management recommendations

(Linker, et al.). Information in the system's database can be

easily updated with current information on weeds and

herbicides. Also, the system can narrow down the number of

available recommendations for a particular problem and rank

them by order of effectiveness. This greatly reduces

extension agents' and grower' time and effort in searching

through a long list of recommendations.

There were some advantages to using herbicide application

as a grove practice for the prototype system. Herbicide

application is a year round operation that must be carried out

regardless of exogenous factors such as weather conditions.

Also, the number of weeds most commonly associated with

citrus, and the herbicides recommended for use on Florida








61
citrus are limited in number. Some factors such as soil type,

severity of weed infestation, and in some cases grove location

and rainfall, influence the herbicide application operation.

Nagarajan et al. (1987) say that weed management involves a

number of strategic and technical decisions, such as chemical

control (pre-plant, pre or post emergence) with a choice of

chemicals and formulations. But once these factors were

identified, almost ideal conditions existed for the

development of an expert system based program.



Program Features

The SIMON concept was used in the development of the

prototype herbicide program to integrate management

recommendations with costs for herbicide application. The

program enables the user to abstract an appropriate herbicide

recommendation from an information database, given a

particular weed problem. A cost information summary is then

given to estimate the expenses that may be incurred if the

recommendation is followed. The program also serves as a

directory, providing information on weeds controlled by each

herbicide. Another feature of the program allows the user to

compare the effectiveness and related costs of using several

herbicides on a particular weed problem.









62

Program Development Environment

SIMON consists of two segments; a herbicide program which

includes the user interface, and an expert system which

determines the correct recommendations for herbicide

application. The following section discusses the programming

environment used for development.

The main herbicide program, like COINS, was written using

dBXL's own application development language. Its design

targets the novice user with little or no computer experience.

To run the program, it is only necessary for the user to

follow instructions in a window at the bottom of each screen.



Expert System Environment

Expert System Shells. Expert systems are usually

developed using a programming environment called a "shell".

Although expert systems can be written in any language, it is

often easier and more efficient to use an expert system shell.

Shells provide comprehensive environments for the development

of rule based expert systems. Expert systems developed using

shells utilize an elaborate process of deductive reasoning

called an inference engine to arrive at a conclusion from a

set of circumstances.

In the case of the expert system portion of SIMON, using

a shell would have proved to be somewhat disadvantageous. The

program was intended to be distributed to growers in Florida

along with the complete COINS system. Most software companies









63
do not offer unrestricted distribution rights for applications

developed using their expert system shells. It is often the

case that costly licenses must be purchased for this purpose.

Another disadvantage to using shells was that expert

systems must be run through the development environment with

which they were written. This increases memory requirements

of the programs being run. Also, expert systems usually

interact with other supporting programs in order to swap

information, and gain access to functions not supported by the

shell used for development. For the herbicide expert system,

a database environment would have been needed to store cost

information and make data queries. Running the expert system

separately from its supporting programs would have

necessitated the exchange of large amounts of data between the

two systems causing a decrease in program efficiency, and

requiring an additional amount of memory.

Maintenance of the expert system was a very important

factor to consider. Love (1988) discusses that maintenance of

the expert system may be quite different from that of the

decision support system. If the expertise changes, whether

evolutionary or revolutionary, the developers of an expert

system must consider methods to help assume that the system's

inference is timely. He further states that while the need to

provide expert system maintenance is important for all expert

systems, it may be particularly appropriate to economics

related expert systems. The changing nature of herbicide









64
formulations, recommendations, and effectiveness on weeds,

necessitated high maintenance requirements, and meant that

herbicide and weed data needed to be readily accessible, and

easily modifiable. It was therefore necessary to ensure that

the rules used by the expert system to arrive at

recommendations, and the recommendations themselves, were not

embedded in the program code. Expert systems developed with

shells usually contain all rules and recommendation within the

program code, and hence, any changes that need to be made

would require the program code to be rewritten and recompiled.

This process placed the use of shells at a disadvantage due

their inefficient maintenance requirements.

A custom designed expert system environment was used for

SIMON. The environment utilizes a unique method of rule

storage and inferencing, whereby rules are stored as records

in a database. The concept involved is fairly simple, and is

flexible enough to be applied to virtually any application

requiring a simple rule based expert system.

Environment Advantages. The environment used for SIMON

was developed using dBXL, allowing the expert system to be

compiled and run in conjunction with the rest of SIMON and the

main COINS program. This keeps memory requirements down, and

allows license-free distribution of the program. The expert

system consists of an algorithm that carries out the

simplified inferencing and deduction process, and database

files containing rules and recommendations. The two are








65
completely separate, allowing modifications to be made to the

data files independently of the program code. As new or

changed information on weeds or herbicides is made available,

the data files can be changed by using the dBXL database

environment, and requires no revisions to be made to the

program code.

Since the expert system part of SIMON is written in dBXL,

it is able to take full advantage of dBXL's functions and

ability to efficiently and easily manipulate large amounts of

information.



Expert System Rule Base

dBXL database files consist of a series of fields and

records. Much like a spreadsheet, fields and records can be

thought of as columns and rows of information. Rules are

stored as records in the database. Each record consists of

three main fields (Figure 5-1); rule number, condition, and

recommendation fields. Rule numbers for each herbicide are

divided into one or more levels of detail allowing easy

referencing. The first level of rule numbers is a whole

number. Rules with decimal values are the next level rules

for a particular condition. File formats for the herbicide

and weed guide are shown in Appendix A.

The deduction process involves an iterative process of

search and selection until a recommendation is found. First,

the expert system displays all condition fields for records

















NUMBER CONDITION

19.00 Areas receiving more than 20"
average annual rainfall

19.00 Areas receiving less than 20"
average annual rainfall

19.00 WEED G-H,

19.10 New plantings.

19.10 Non-bearing established plantings.

19.11 Soil texture is coarse.

19.11 Soil texture is medium

19.11 Soil texture is fine

19.20 New plantings.

19.20 Non-bearing established plantings.

19.21 Soil texture is coarse.

19.21 Soil texture is medium.
or Soil texture is fine with 2-5%
organic matter.

19.21 Soil texture is fine.
or Soil has 5-10% organic matter.

19.21 Soil has 2-5% organic matter.

19.21 Soil has 5-10% organic matter.


RECOMMENDATION


NEXT 19.10


NEXT 19.20

Recommendation

NEXT 19.11

Recommendation

Recommendation

Recommendation

Recommendation

NEXT 19.21

Recommendation

Recommendation



Recommendation


Recommendation

Recommendation

Recommendation


Example Rules Used for the Herbicide Treflan


Figure 5-1










with first level rule numbers. The user is asked to choose a

condition from the list. The recommendation field for the

chosen condition will either direct the program to search for

another set of conditions, or will contain a recommendation.

In the case of the latter, the program terminates the search,

and displays the recommendation. If a second set of

conditions exist, they are displayed based on their rule

numbers, and the user is asked to choose again. This

procedure is repeated until the program comes across a

recommendation. The program can also accommodate special

cases for rules where only a particular situation within a set

of rules requires a specific recommendation. This is the case

when a particular weed requires a specific or unique

recommendation. A flow chart illustrating this deduction

process is shown in Figure 5-2.



Knowledge Acquisition

Identifying the Experts

The methods by which the program arrives at management

recommendations are based on the thought processes used by

experts in the field of weed science and particularly in the

area of citrus production. The first step in the design of

the program was to identify the experts.

Two citrus herbicide experts were identified. Dr. Megh

Singh and Dr. David Tucker are both researchers at the

University of Florida's Citrus Research and Education Center











Identify first level
rule NUMBER for the
recommended herbicide

One weed
selected


Check all the
herbicide's
first level
conditions for
the selected
WEED
I


Several weeds
selected


Display all first
level conditions
for the herbicide


Select the
appropriate
condition


Check the
RECOMMENDATION
for the selected
rule


No


Display the
recommendation
found


Figure 5-2


RECOMMENDATION
prompts for NEXT
rule NUMBER


Yes


Advance to, and
display the
next level of
conditions


Flow Chart Illustrating the Deduction Process
Used by SIMON





Special rules
found for the
selected weed


Yes No


I I


Il


ij









69

in Lake Alfred, Florida. Both experts had worked closely with

citrus growers and extension agents, and were familiar with

situations that may arise concerning weed infestation in

citrus. They were also current on developments in the

herbicide industry, and hence they provided the complete

knowledge required to develop the system. The herbicide and

weed guide also relies heavily on published materials for

information. However, even with the immense amount of

information available in the form of extension guidelines, the

actual application of this "domain knowledge" to specific

situations had been (and still is) provided primarily by weed

specialists (Holt, 1988).


Preliminary Interview

Determining how the experts arrived at a conclusion from

a series of circumstances was the next logical step in the

program design. Due to the distance between the programmer

and the experts, personal interviews were limited to a

preliminary interview and several follow up interviews. All

other consultations were made over the phone, and through the

Institute of Food and Agricultural Sciences' (IFAS) electronic

mail system. Beck et al. (1987) mention that in general,

extension specialists cannot spend significant portions of

their time working on an expert system project. Furthermore

the experts and engineers are typically quite distant

geographically and cannot meet for intensive interviews on a









70

regular basis. The purpose of the preliminary interview was

to allow the experts to discuss freely the pertinent

considerations of the system (Lacey et al., 1989).

During a preliminary interview the experts were given a

general idea of the concepts behind COINS. A discussion

followed on the steps involved in the development of SIMON,

and the herbicide expert system. Requirements and limitations

of the herbicide expert system were also discussed. It was

determined that only around twenty herbicides were recommended

by IFAS for use on Florida citrus. These herbicides were used

as the basis for the system's herbicide database. It was also

decided that only a certain number of weeds commonly occurred

on Florida citrus, and that the weed database should be

limited to these weeds.



Sources of Information

Information on weed names and classifications were

derived from chemical manufacturers' publications as well as

independent studies conducted by the experts and the herbicide

manufacturers. A list of herbicides recommended for use on

citrus was obtained from the 1990 Citrus Spray Guide (IFAS

Publications, 1990).

The expert system rules used to arrive at the proper

recommendations were derived from several sources. The main

source of the spray recommendations, including rates,

scheduling, and other information, was the 1989 Crop









71

Protection and Chemicals Reference (CPCR). The CPCR contained

product labels for all the recommended herbicides. Product

labels contain all necessary information needed to use each

product, and are used by the grower as a reference.

Additional spray recommendations were obtained from the Citrus

Spray Guide.



Establishing the Decision Process

It was necessary to establish the thought process that

the experts used to analyze a particular weed problem.

Following a discussion with the experts a decision tree was

drawn up that describes the logic used (Figure 5-3).

The first step involved the definition of a weed problem.

Categories as well as names of weeds were selected. Weeds in

three categories, namely grasses, broadleaf weeds, and vines,

were selected to constitute a weed problem. Weed categories

and weed names are listed in Tables 5-la, 5-lb, 5-1c. The

next step was to identify what herbicides control the weeds

selected. This is done by referring to the herbicide product

labels or the citrus spray guide. If no herbicides were found

to control all weeds, the next best herbicides that control as

many weeds as possible were selected.

When the herbicides identification process is completed,

the herbicides were ranked as to their effectiveness on the

weeds selected. The ranking is based on a susceptibility

table that was developed by the experts in conjunction with








72
the chemical manufacturers. It shows the effectiveness of

each herbicide on weeds in each category (Table 5-2). The

following is an explanation of the five susceptibilities

listed in descending order of effectiveness.

S indicates that weeds are susceptible to the herbicide

at germination, and at early seedling stage of growth. Weeds

may also be susceptible at later stages of maturity.

PS indicates that weeds are susceptible only at

germination. Repeat applications may sometimes control

established weeds.

I denotes intermediate control indicating that the degree

of control will be erratic with some plants within a species

population being killed and others not.

T indicates that weeds are tolerant and either showing no

signs of injury or able to recover from injury symptoms. U

indicates that the susceptibility status is unknown due to

lack of experimental data and reliable field observations.

In order to rate the herbicides' effectiveness on a

combination of weeds, a rating system was established by the

experts and the knowledge engineer. The S, PS, I, T, and U

susceptibilities were given equivalent numerical ratings on a

scale of one to five, with five being the highest rating

corresponding to a susceptibility of S. The ratings for all

the weeds selected were summed and resulted in an overall

rating for each herbicide.












Define weed problem


Select weed Category


Select weeds


Identify the herbicide
or herbicides that control
the weed problem selected
I I-

One or more herbicides -- Find the next
are found to control No best herbicide
all weeds selected


Yes
Rank the herbicides
by susceptibility of
the weeds controlled


Classify herbicides as
premergence,
postemergence, or both



List of available
herbicides is -- No
satisfactory

Select a herbicide
Yes from the list and
get recommendation
and cost summary



Recommendation
and costs are No
satisfactory


Yes End


Flow Chart for Experts' Decision Process


Figure 5-3























Table 5-la


List of Weeds by Category. a) grass weeds, b)
broadleaf weeds, c) vines


Weed Name Scientific Name


Bahaigrass
Bermudagrass
Carpetgrass
Cattail
Crabgrass
Crowfootgrass
Goosegrass
Guineagrass
Johnsongrass
Maidencane
Napiergrass
Natalgrass
Nutsedge
Pangolagrass
Paragrass
Peppergrass
Sandspur
Signalgrass
Texas Panicum
Torpedograss
Vaseygrass
Yellow Foxtail


Paspalum notatum
Cynodon dactylon
Axonopus affinis
Typha sp.
Digitaria adscendens
Dactyloctenium aegyptium
Eleusine indica
Panicum maximum
Sorghum halepense
Panicum hemitomon
Pennisetum purpureum
Rhynchelytrum repens
Cyperus rotundus
Digitaria decumbens
Panicum purpurascens
Lepidium virginicum
Cenchrus echinatus
Brachiaria piligera
Panicum Texanum
Panicum repens
Paspalum urvillei
Setaria glauca














Table 5-lb


List of Weeds by Category. a) grass weeds, b)
broadleaf weeds, c) vines


Weed Name


Scientific Name


Bitter Mint
Black Nightshade
Brazilian Pepper
Camphorweed
Ceaserweed
Common Purslane
Common Ragweed
Creeping Charlie
Cudweed
Dayflower
Dogfennel
Evening Primrose
Flat-topped Goldenrod
Florida Beggarweed
Florida Pusley
Goatweed
Goldenrod
Horseweed
Jerusalem Oak
Lambsquarters
Lantana
Mexican Tea
Pepperweed
Pigweed
Pokeberry
Primrose Willow
Rouge Plant
Rustweed
Saltbush
Seamyrtle
Skunkweed
Sowthistle
Spanish Needles
Spurge
Swampwillow
Teaweed
Virginia Pepperweed
Waxmyrtle


Hyptis mutabilis
Solanum Nigrum
Schinus terebinthifolius
Heterotheca subaxillaris
Urena lobata
Portulaca oleracea
Ambrosia artemisiifolia
Lippia nodiflora
Gnaphalium sp.
Commelina benghalensis
Eupatorium capillifolium
Oenethora sp.
Euthamia minor
Desmodium tortuosum
Richardia scabra
Scoparia dulcis
Solidago sp.
Conyza canadensis
Chenopodium botrys
Chenopodium album
Lantana camera
Chenopodium ambrosioides
Lepidium virginicum
Amaranthus sp.
Phytolacca americana
Ludwigia peruviana
Rivina humilis
Polypremum procumbens
Baccharis halimifolia
Baccharis halimifolia
Achyranthes aspera
Sonchus sp.
Bidens pilosa
Chamaesyce hyssopifolia
Salix nigra
Sida acuta
Lepidium verginicum
Myrica cerifera


-----------------------------~----~----- -----------























Table 5-lc


List of Weeds by Category. a) grass weeds, b)
broadleaf weeds, c) vines


Weed Name


Scientific Name


Air Potato
Balsam Apple Vine
Bigroot Morningglory
Brazilian Nightshade
Briars
Calico Vine
Cats Claw Vine
Cypress Vine
Maypop (Passion Flower)
Milkweed (Strangler) Vine
Moonvine
Morningglory
Narrow-Leaf Milkweed Vine
Peppervine
Rosary Pea
Virginia Creeper

Wild Grape
Wild Watermelon (Citron)
Woevine


Dioscorea bulbifera
Momordica charantia
Ipomoea pandurata
Solanum seaforthianum
Smilax sp.
Aristolochia littoralis
Bignonia unguis-cati
Ipomoea quamocilt
Passiflora incarnate
Morrenia odorata
Ipomoea alba
Ipomoea sp.
Cynanchum scoparium
Ampelopsis arborea
Abrus precatorius
Parthenocissus
quinquefolia
Vitis rotundifolia
Citrullus vulgaris
Cassytha filiformis


------------------------------------------------------------











Weed Susceptibilities


Herbicide Codesa
Weed Name A B C E F G H I J


Bahaiagrass
Bermudagrass
Common Carpetgrass
Cattail
Crabgrass
Crowfootgrass
Goosegrass
Guineagrass
Johnsongrass
Maidencane
Napiergrass
Natalgrass
Purple Nutsedge
Pangolagrass
Paragrass
Southern Sandspur
Hairy Signalgrass
Texas Panicum
Torpedograss
Vaseygrass
Yellow Foxtail
Peppergrass
Florida Beggarweed
Bitter Mint
Black Nightshade
Brazilian Pepper
Ceasarweed
Camphorweed
Common Purslane
Common Ragweed
Creeping Charlie
Cudweed
Dayflower
Dogfennel
Evening Primrose


S S S I U
PS PS PS T U
S S S I U
U U U U U
S S S S U
S S S S U
S S S S U
PS PS PS PS U
PS PS PS PS U
PS PS PS PS U
T T T T U
S S S PS U
I I I T U
S S S I U
S S S I U
S S S U
PS PS PS I U
S S S U U
PS PS PS T U
PS PS PS T U
S S S S U
U U U U U
S S S U U
S S S U U
S S S S U
U U U U U
U S S U U
S S S S U
S S S S U
S S S PS U
S S S U U
S S S U U
S S S S U
PS PS PS PS U
U U U U U


T S T U
S S T U
U S T U
U S T U
S S S S
PS S S U
S S S S
S S S U
S S PS PS
U S U U
S S U U
U S U U
T PS PS U
U S T U
S S PS U
S S PS U
S S PS U
S S PS U
I PS T U
S S T U
S S PS I
U U U U
T S T U
T S U U
T S PS U
T I T U
T S U U
T S U U
T I T S
T S PS I
T I T U
T S PS S
T PS T U
T S T U
T S T U


a. The letters correspond to the following herbicides


A. Bromacil
B. Bromacil & Diuron
C. Bromacil & Diuron
E. Diuron


F. EPTC
G. Fluazifop-Butyl
H. Glyphosate
I. Metolchlor
J. Napropamide


Table 5-2











Table 5-2--Continued


Herbicide Codesa
A B C E F G H I J


Weed Name


Florida Pusley
Flat-Topped Goldenrod
Goatweed
Goldenrod
Horseweed
Jerusalem Oak
Lambsquarters
Lantana
Mexican Tea
Pepperweed
Pigweed
Pokeberry
Primrose Willow
Rouge Plant
Rustweed
Saltbush
Seamyrtle
Skunkweed
Sowthistle
Spanish Needles
Spurge
Swampwillow
Teaweed
Virginia Pepperweed
Waxmyrtle
Air Potato
Balsam Apple Vine
Bigroot Morningglory
Brazilian Nightshade
Briars


S S S PS U
PS PS PS PS U
T S PS PS U
PS PS PS PS U
U U U U U
PS PS PS I U
S S S PS U
PS PS PS PS U
U U U U U
S S S S U
T PS S S U
S S S PS U
U S S U U
S S S PS U
S S S U U
T T T T U
U U U U U
PS PS PS PS U
U S S U U
T S PS S U
S S S PS U
U U U U U
PS S PS PS U
S S S U U
U U U U U
T T T T U
PS S S PS U
T U T U U
U U U U U
T T T T U


a. The letters correspond to the following herbicides


A. Bromacil
B. Bromacil & Diuron
C. Bromacil & Diuron
E. Diuron


F. EPTC
G. Fluazifop-Butyl
H. Glyphosate
I. Metolchlor
J. Napropamide


S T I
S T U
I U U
S T U
S U U
PS T U
S PS S
S T U
U U U
S T U
S PS S
S U U
S U U
U U U
U U U
S T U
S T U
U U U
S T U
S T U
S T U
S T U
PS T U
S PS U
I T U
PS T U
S T U
PS T U
S T U
I T U


-- -- -- -- -- - - -- -- --






















Table 5-2--Continued


Herbicide Codesa
A B C E F G H I J


Weed Name


Calico Vine
Cat'S Claw Vine
Cypress Vine
Maypop
Milkweed (Strangler) Vine
Moonvine
Morningglory
Narrow-Leaf Milkweed Vine
Peppervine
Rosarypea
Virginia Creeper
Wild Grape
Wild Watermelon
Woevine


T T T T U
U U U U U
PS PS PS I U
PS PS PS PS U
I I I T U
PS PS PS PS U
T T T T U
T T T T U
PS PS PS PS U
T T T T U
T T T T U
U PS PS U U
T T T T U


a. The letters correspond to the following herbicides


A. Bromacil
B. Bromacil & Diuron
C. Bromacil & Diuron
E. Diuron


F. EPTC
G. Fluazifop-Butyl
H. Glyphosate
I. Metolchlor
J. Napropamide


I T
PS T
S T
S T
I T
UT
S T
I T
I T
S T
PS T
I T
S T
U T













Table 5-2--Continued


Herbicide Codesa
Weed Name K L M N O P Q T
---------------------------------------------mm-----------


Bahaiagrass
Bermudagrass
Common Carpetgrass
Cattail
Crabgrass
Crowfootgrass
Goosegrass
Guineagrass
Johnsongrass
Maidencane
Napiergrass
Natalgrass
Purple Nutsedge
Pangolagrass
Paragrass
Southern Sandspur
Hairy Signalgrass
Texas Panicum
Torpedograss
Vaseygrass
Yellow Foxtail
Peppergrass
Florida Beggarweed
Bitter Mint
Black Nightshade
Brazilian Pepper
Ceasarweed
Camphorweed
Common Purslane
Common Ragweed
Creeping Charlie
Cudweed
Dayflower
Dogfennel
Evening Primrose

a. The letters correspond to the


K. Norflurazon
L. Oryzalin
M. Oxyfluorfen
N. Paraquat Dichloride


PS PS T I PS U T PS
PS PS T I PS I T PS
I S U I S U T S
PST U S U U T T
S S I PS S U S S
PS S I PS S U S S
PS S I S S S S S
PS S I S S U PS S
PS PS I PS S PS PS PS
PST U T U U T T
PS PS U I PS U T PS
PS PS I U PS U I PS
I T I T T U T T
PS PS I PS PS U T PS
PS PS U T PS U T PS
PS S I S S U I S
PS S U S S U PS S
PS S I S S S I S
S PS I T PS U T PS
PS PS U PS PS U T PS
PS S I S S S S S
U U U U U U U U
PS T I S T U S T
I T U U U U U T
PS T S I T U S T
U T U I T U T T
U T U U U U U T
I T PS PS T U U T
S S S S S U S S
PST I S T U S T
U T I S T U S T
PS T PS S T U S T
PS T I S T U T T
S T U PS T U PS T
I T PST T U I T


following herbicides


O. Pendimethalin
P. Sethoxydim
Q. Simazine
T. Trifluralin















Table 5-2--Continued


Herbicide Codes'
Weed Name K L M N O P Q T

Florida Pusley PS S S S S U I S
Flat-Topped Goldenrod PS T U U T U U T
Goatweed PST U U T U I T
Goldenrod PS T PS T T U S T
Horseweed I T T PST U I T
Jerusalem Oak I T PS PS T U T T
Lambsquarters PS S S S S U S S
Lantana T T T I T U T T
Mexican Tea U U U U U U U U
Pepperweed PS T S PS T U I T
Pigweed I S S S S U S S
Pokeberry PS T U PS T U S T
Primrose Willow PS T U PS U U U T
Rouge Plant U T U U U U U T
Rustweed I T U U U U U T
Saltbush I T PS PS T U T T
Seamyrtle I T U PS U U T T
Skunkweed U T U U U U U T
Sowthistle I T S S T U S T
Spanish Needles I T PS S T U S T
Spurge S S PS S S U S S
Swampwillow U T PS PS T U T T
Teaweed PST S T T U I T
Virginia Pepperweed PS S S PS U U I S
Waxmyrtle U T U PST U T T
Air Potato U T U U U U T T
Balsam Apple Vine PS T PS PS T U S T
Bigroot Morningglory I T U PS T U S T
Brazilian Nightshade I T U T T U S T
Briars T T U PST U T T

a. The letters correspond to the following herbicides


K. Norflurazon
L. Oryzalin
M. Oxyfluorfen
N. Paraquat Dichloride


O. Pendimethalin
P. Sethoxydim
Q. Simazine
T. Trifluralin























Table 5-2--Continued


Herbicide Codesa
K L M N O P Q T


Weed Name


----------------------------------------------------------
Calico Vine U T U PS U U T T
Cat'S Claw Vine U T U T U U T T
Cypress Vine I T PS PS T U T T
Maypop T T PS PS T U T T
Milkweed (Strangler) Vine PS T PS PS T U PS T
Moonvine U T PS PS U U T T
Morningglory PS T S PS T U I T
Narrow-Leaf Milkweed Vine U T U PS T U U T
Peppervine U T U PS U U T T
Rosarypea U T U U U U T T
Virginia Creeper U T U PS T U T T
Wild Grape T T U PST U T T
Wild Watermelon I T PS S T U T T
Woevine T T U U U U U T
a. The letters correspond to the following herbicides
a. The letters correspond to the following herbicides


K. Norflurazon
L. Oryzalin
M. Oxyfluorfen
N. Paraquat Dichloride


O. Pendimethalin
P. Sethoxydim
Q. Simazine
T. Trifluralin








83

Apart from the susceptibility ratings, herbicides were

classified into three categories. Pre-emergence herbicides

which are used prior to weed emergence to control potential

problems. Post-emergence herbicides which are used to control

established weeds. And lastly herbicides that offer both pre-

emergence and post-emergence control. The classification is

used to determine which types of herbicides are suited for a

particular situation.

After rating and classifying the herbicides, a selection

among alternatives was made. The experts agreed that once all

recommended herbicides have been presented to the user, it is

up to that user to choose which herbicide to utilize. They

were not willing to specify a particular herbicide to a grower

to avoid showing any type of support or endorsement for one

product.

Following the selection of a herbicide, recommendations

for spraying rates were given based on information found in

the product labels. In most cases more than one

recommendation was listed depending on such factors as types

of weeds, soil moisture conditions, and soil type. The final

recommendation depended on these factors. Spraying cost

estimates were then given based on recommended spray rates,

product costs, and treated acreage.










Follow-up Interviews

Two follow-up interviews were conducted throughout the

program development process. The program was evaluated by the

experts at each follow-up interview with special emphasis on

the new features added since the last interview. Ideas on

improvements and enhancements were exchanged. During the

final follow-up interview, a complete list of program features

was established, and updates on herbicide and weeds

information were made. Once all enhancements and features in

the list were carried out, the final version of the program

was ready for evaluation.



Program Execution

The program proceeds in one of three ways; a herbicide

recommendation is given based on a weed problem, a list of

weeds controlled by a herbicide is identified, or two or more

herbicides are compared as to their effectiveness on a weed

problem.



Obtaining a Herbicide Recommendation

The user defines a weed problem by choosing weeds from

lists in the grass, broadleaf, and vine categories. The

program searches for herbicides that control the weeds

selected. If one or more herbicides are found to control all

weeds selected, they are ranked and displayed on the screen

(Figure 5-4). If no herbicides are found to control all weeds











Herbicide and Weed Guide


Available Herbicides for All Weeds Selected


Trade Name Common Name Rating Effect.

Roundup Glyphosate 6 Post
Devrinol Napropamide 6 Pre
Solicam Norflurazon 7 Pre
Goal Oxyfluorfen 6 Both
Gramoxone Paraquat Dichloride 8 Post


Susceptibility Rating The susceptibility rating
is a measure of the effectiveness of each herbicide
on the weed combination it controls. It is the sum
of the susceptibiltities of each weed to the
herbicide.


Figure 5-4


Example List of Herbicides Available to
Control The Defined Weed Problem























Herbicide and Weed Guide


Note No herbicides were found that control
all weeds selected.

The program can now proceed in two different ways:

1 The program will search for herbicides that control
as many weeds as possible from the list of selected
weeds. One herbicides is selected from a list, and
recommendations and spraying information are given
for that herbicide.

2 The program will search for individual herbicides
for grass weeds, broadleaf weeds, and vines.
Herbicides that control as many weeds as possible
in each category are found. A herbicide is selected
for each category, and recommendations and spraying
information are given for each herbicide.




Figure 5-5 Option Screen When One Herbicide is Not Found
to Control All the Weeds Selected









87

selected, the user is prompted to direct the program in one of

two ways (Figure 5-5). The program can search for the best

herbicide for all weeds selected, or the best herbicides for

weeds in each category. Once herbicides have been identified,

the user may either select a herbicide, or reject the choices,

and direct the program to search for the next best

alternatives. The search continues until the user accepts a

herbicide and requests a recommendation. If herbicides are

recommended separately for each category of weeds, a

combination of two or more herbicides must be used. The

program only lists these herbicides and the individual spray

rates for each. It does not give information on mixing

procedures, compatibility, and precautions. The user must

refer to more detailed information by consulting the product

labels.

Recommendations for all herbicides are stored in a

database, and with the aid of the expert system and some input

from the user, the appropriate recommendation for the selected

herbicide is given (Figure 5-6). At this point in the

program, a cost estimate for the selected herbicide can be

obtained. The total cost for spraying the herbicide at the

recommended rate reflects the material costs for using the

herbicide. Other costs are usually incurred during the

operation such as machinery, labor, and management costs. The

cost per unit of purchasing the herbicide and number of

treated acres to be sprayed are required from the user. If






















Ryvar I (Bromacil) for use on Texas Panicum



Recommendation

Apply 4-5 lbs. of HYVAR X per acre during the
period from winter to early summer. Alternatively,
make two applications of 3-4 lbs. per acre per
year in spring and summer. Partial control usually
occurs with a single treatment; repeat applications
are required to control perennial weeds. Control
of perennial weeds may be improved by cultivation
prior to treatment; otherwise, avoid working the
soil as long as weed control continues since
effectiveness may be reduced.


Susceptibility Susceptible at germination and
early seedling stage of growth. May also be
susceptible at later stages of maturity.


Figure 5-6


Example Recommendation Screen for Using Hyvar
X Herbicide on Texas Panicum



























Herbicide Cost Information Krovar I


Cost per gallon for this herbicide ($) 7.35
Number of treated acres to be sprayed 5.00

Rates in pounds per treated acre -
Recommended range 2.00 4.00
Recommended maximum 8.00

Total Cost ($) Range per treated acre 14.70 29.40
Range for this grove 73.50 147.00

Maximum Cost ($) Per treated acre 58.80
For this grove 294.00


Figure 5-7


Example Cost Screen for Krovar I Herbicide









90

additional spray material such as oils or surfactants are to

be used, the user must specify the cost of these materials.

When all information is provided, the program calculates the

cost per acre and total grove cost for the herbicide based on

the recommendation and the user's inputs (Figure 5-7).



Obtaining a List of Weeds Controlled by a Herbicide

The user chooses a herbicide from a list of all

herbicides in the database (Table 5-3). The program searches

for weeds controlled by that herbicide and provides lists of

weeds in the grass, broadleaf, and vine categories. The user

may select any weed from the lists and obtain appropriate

spraying recommendations and cost information.



ComDarina Several Herbicides

It is often the case that a grower has several choices of

herbicides to use on his weed problem. A is very difficult to

assess the advantages of using one herbicide over another. An

accurate management decision could lead to a reduction in

costs of spraying while maintaining an acceptable level of

weed control. A feature of the program allows a comparison to

be made between several herbicides resulting in a better

management decision. Their effectiveness on a particular weed

problem as well as the cost of using each herbicide are

compared. After selecting the herbicides from a list, and

defining a weed problem, a summary screen is displayed for