Computers and Industrial Engineering Vol. 23, Nos I--4, pp. 365-367, 1992 0360-8352/92 $5.00+0.00 Printed in Great Britain Pergamon Press Ltd

EXPERT SYSTEM FOR IE STUDENT ADVISEMENT

Abel Fernandez, John Biegel, Jeff Earhart

Department of Industrial Engineering and Management Sciences University of Central Florida

Orlando, FL 32816

ABSTRACT

Student advising is a key function within an academic environment. An expert system may be used as a first line of response to students' advising needs. The system currently under development, uses hypermedia as the interface between user and hidden expert system routines. Hypermedia allows use of user oriented facilities (icon driven menus, branching following user requests, digitized photographs, video discs, animation, etc). The software architecture chosen is that of a hypergraph with embedded, distributed expert systems.

PURPOSE OF PROGRAM

Student advising is an indispensable and key function of every academic department. A properly administered academic advising program develops a long term faculty-student relationship. However, expert systems may be useful as a first line of response to students' advising needs. This is particularly important if one considers that the majority of students are self-advised. Most students seek formal advisement only if they experiencing problems - a very reactive process. One objective of the expert advising system is increase proaetive use of advisin&

Expert systems make use of specialized knowledge to solve problems to an equivalent level of a human expert. In this particular context, the knowledge based system has been designed to supplement the role of the expert faculty advisor in recommending course sequences. However, a second role of a faculty advisor is to provide counsel on University policy and rules. Information is available to the student through a maze of printed sources - catalogs, student handbooks, memos, etc. In this second role, the advising system provides a source of information retrieval.

PLATFORM SELECTION

A critical decision in the implementation of an advising system is the inference engine to be used. Decision criteria to be considered include: cost, portability, availability of student assistance, accessibility, maintainability, available development tools, and user interface capabilities.

In our academic context the critical criteria were deemed to be cost, accessibility of software and documentation, and availability of students for development work. Considering these factors as crucial, and overshadowing other considerations, the expert system language chosen wasClips. Clips (C Language Integrated Production System) VERSION 4.2, is a forward chaining rule-based language developed at NASA/Johnson Space Center. It is available at nominal cost through COSMIC (University of Georgia), has good documentation (it is the basis for the text "Expert Systems - Principles and Programming" and has NASA provided user documentation) and is taught in a graduate level course within Industrial Engineering. The latter provides a good source of student assistance for development work.

Although a very flexible and relatively "easy" language, Clips does not offer the powerful tools of fourth generational software environments. In particular, it does not directly provide a user-friendly human interface. Developers are required to program the interface through another program language (FORTRAN, C, Ada, etc.), calling the Clips subroutines as needed.

development handicap has been overcome through the recent availability of Hyperclips, available at nominal ec~t through COSMIC. HyperClips integrates the powerful front-end and information retrieval capabilities of HyperCard CI'M), with the inference engine capabilities of Clips. HyperC~rd stacks may call Clips subroutines to provide problem solving and analysis capabilities. An ideal combination for a student advising system, or similar user and information intensive applications.

365

366 Proceedings of the 14th Annual Conference on Computers and Industrial Engineering

HyperCard is an information storage and retrieval system which exploits the graphics and user interface capabilities of the Apple Macintosh (TM) computer. It may be viewed as a powerful development platform for information based applications, allowing rapid prototyping yet providing sophisticated presentation techniques. Hypermedia, such as the HyperCard software, is becoming widely available and understood, and allows the use of user oriented facilities such as icon driven menus, branching following user requests, digitized photographs, video discs, animation, etc.

Although hypermedia's powerful information storage and retrieval capabilities were potentially attractive, it was important to ensure that we were not fitting the problem to the available tool In 'The Society of Text" the following are suggested criteria for using hypermedia in a particular application: large body of separable knowledge; interrelated components; and, user need of limited knowledge chunks. The student advising problem meets these criteria and is potentially a good application for a hypermedia solution.

PROGRAM STRUCTURE - CONCEPTUAL FRAMEWORK

An important development stage is the determination of the user tasks. What is the user seeking and how is the information best structured in order to facilitate task accomplishment? The system should strive to provide search and information retrieval which is accurate, complete and comfortable for the end user. A hypermedia system may be enhanced by an expert system so as to: 1) help the user select an optimum path to desired information; 2) provide expert "opinion" on recommended course of action (viz, courses to take during a particular semester).

There are a number of possible software architectures that may be used to integrate expert systems and hypermedia. The approach considered most appropriate for this application is the creation of a hypertext with embedded, distributed expert systems. This distributed system, illustrated in Figure 1, consists of interlinked hypermedia nodes with embedded independent, small knowledge bases distributed throughout the nodes. (Hypermedia may be diagrammed as hypergraphs - nodal pools of information with intereormecting links between them.) This modular approach provides significant advantages and benefits.

Flexible use of knowledae bases The knowledge base at a particular node may be adapted to its particular use at that node. It may be used as a navigational aid to guide the user to information or to provide expert counsel.

Facilitate developmen¢ Module segmentation of expert system facilitates program development since routines may be independently conceptualized, developed, tested and debugged.

Flexible use of hvvermedia nresentation resources Modules may independently make use of appropriate hypermedia input/output facilities.

Facilitate oro~ram maintenance The independent knowledge bases are smaller and compartmentalized, thus facilitating- data maintenance.

Facilitates oro~am expansion Future independent modules may be added as needed, with links to existing modules proviciing network connections.

SYSTEM NETWORK

Leveling techniques may be used to present the network at progressively greater levels of detail. At the highest level, the nodes shown in Figure 1 represent modules of information on the Florida State University System, the University, the College of Engineering, system help and other similar macro level chunks. Links allow the user to move from node to node, with internal expert systems at each node to provide "opinion" and/or navigational assistance. As diagrams move to lower levels the nodes in Figure 1 represent greater detail: for example, the nodes within the College of Engineering, Again, linkages provide mobility, and localized expert system routines provide as-needed assistance. This leveling technique may be used to continue showing greater granularity within given nodes of the hypergraphs.

HyperCard graphics capabilities are used to provide effective input/output user interfaces. Icons, point and click selections, pop up menus, input validity checking, and other HyperC.ard features are used to facilitate user entries. In addition, wherever poss~le students are presented with data screens that replicate familiar paper forms and flow charts. It should also be noted that HyperCard offers more than powerful interface and information retrieval capabilities. Through its HyperScript programming language, it also provides number processing and logic features. These, for example, will be used to perform GPA calculations, tabulate credits taken, etc. The logic features may be used to perform such functions as data validity checking, predicated function processing (allow movement to another screen only if a certain field has been entered), and other similar functions.

FERNANDEZ et al.: Student Advisement 367

FI6UBE 1 - ADVTSING SYSTEM HYPERGRAPH

PRESENT STAGE OF DEVELOPMENT

Development of the described system began in January of this year. Prior to that date, a prototype system written in Clips was developed to examine the feasibility of using an knowledge based student advising system. The prototype system demonstrated feasibility. The challenges of the present development work are to devise an overall system design that transcends a narrow definition of student advising. A true and well rounded "advisor" provides counsel and information not only in course sequences, but also in other academic and university areas. The present design envisions an ultimate goat of a cohesive and comprehensive information retrieval and counselling system. However, in approaching this ultimate goal, the following questions must be addressed: Can a system be developed that will be accepted and used by students? Do we want to create such a system?

BIBLIOGRAPHY

Barrett, Edward, editor, "The Society of Text", The MIT Press, Cambridge, Massachusetts, 1989

Giarratano, Joseph and Riley, Gary, '~_.xpert Systems - Principles and Practises", PWS KENT Publishing Company, Boston, Massachusetts, 1989.

Harvey, Greg, "Understanding HyperCard", Sybex Inc., Alameda, California, 1988.

Martin, James, "Hyperdocuments and How To Create Them", Prentice Hall, Engiewood Cliffs, New Jersey, 1990.

Taner, Murat; Aksov, Yasemin; and Arroyo, Antonio, "An Expert System for Academic Advising", Applied Artificial Intelligence: 5:267-280, 1991, Hemisphere Publishing Company.

HyperClips User Documentation, COSMIC, The University of Georgia, Athens, Georgia, 1991.