Laboratory Exercises For Computer Science 153

Expert Systems

Goals: This laboratory exercise introduces the concept expert systems in general and a specific expert system in particular. Further, since such programs often utilize file input, this application provides an example of the need to convert data generated by one program into a different form for use in another program.

Background:

Every college and university must decide how to place incoming students in appropriate courses. In many cases, placement recommendations may be based on standardized tests (ACTs, ACTs, Achievement Tests, Advanced Placement (AP) scores, International Baccalaureate (IB) scores) and high school transcripts. Historically at Grinnell, faculty reviewed such data manually and made recommendations.

While faculty continue to be actively involved, the Mathematics and Computer Science now uses an expert system to automate much of this process. The first version of the current program was developed in the Spring, 1993, as a student-faculty project; students Vikram Subramaniam and Ivan Sykes worked with faculty member Henry Walker. The program has been refined and expanded several times over the years to reflect new or revised courses.

Overview of the Placement Program:

The main program can be used in several ways:

users can run the program interactively at a terminal window with the user entering needed data,
the program can process information received electronically from the Registrar, producing letters for students, or
users can use a World-Wide-Web interface to enter data and view placement conclusions.

The Web interface is simplest.

Try running this program through Netscape, using the address http://www.math.grin.edu/~walker/placement/placement-form.html

While the placement program is written in LISP, the code and run-time environment are quite similar to Scheme. To run the program in the relatively familiar dtterm environment, follow these steps:

Run Allegro Common Lisp in a dtterm by typing the command
```
       acl
```

At the new prompt, type

       (load "~walker/placement/newstudents.lsp")
       (placement)

When the menu appears, use the "I" option to run the program interactively. Then, enter data for yourself or for a hypothetical student as requested.
When you are done, type "Q" to quit the menu. Then at the "USER" prompt, type
```
       (exit)
```

This application illustrates the structure of many expert systems. This system has two basic parts:

A rule base, found in file "~walker/placement/newstudents.lsp", which contains about 90 rules concerning the placement of students.
An inference engine, found in file "~walker/placement/tmycin.lsp", which applies the rules to specific cases.

Each part is written in LISP, the most common computer language for artificial intelligence. The inference engine is based on the program TMYCIN, written by Gordon Novak at the Artificial Intelligence Laboratory of the University of Texas at Austin.

To view the rules in a dtterm window, type
```
       more ~walker/placement/newstudents.lsp
```
This command displays the first screen of the rule base. Hit the space bar to view subsequent screens.
While the programming details require considerable sophistication in LISP, you will notice that many elements of the code are reasonably similar to Scheme -- following a list-oriented format. The general organization of the rule base is reasonably straightforward:
- After loading the inference engine ("/users/walker/placement/tmycin.lsp"), the main variables (i.e., the context) for placement is given.
- Semesters of English and high school credits are used to infer how many semesters of high school work are recorded on the current transcript.
- Standardized test scores are classified high, good, fair, poor, and low.
- Preliminary placements are made for mathematics and then for computer science by considering minimum cutoff levels. (Note, as the rules specify minimums, someone prepared for calculus II also will be adequately prepared for calculus I.)
- Final placements are determined as the highest tentative placement.
A certainty factor is associated with each rule, based on a 1000 point scale. A factor of 1000 indicates great confidence with the conclusion, while 700 or 500 shows must less certainty in the result.
Use your knowledge of Scheme/LISP lists to interpret the likely syntax of rules in this system.

Files for the Placement Program:

When this system is used each year to make recommendations to incoming students, the Registrar's Office send the Department of Mathematics and Computer Science a file of student transcript information. A fictional file of this type is available in ~walker/261/labs/student.raw-data.

Examine this data file to determine the type and format of information contained.

Since the expert system is written in LISP, which is list-oriented, the placement program takes input from a file, in which data for a student is in a separate list. For example, a typical entry might be:

("Person I M A" ( (ACT 28) (SemOfEng 8)(SemOfMath 8)
         (Grades 2.75)(SemOfPCalc 1)(PCalcGrades 3.00)
         (SemOfCS 2)(CSGrades 4.00) (TSemUnits 36) ) 
     (Campus-Box |20-01| ) (Adviser "My Friend") )

The full file for the above fictional students is available in ~walker/261/labs/student.data. Here, the person's name is the first entry in the list. The second list component is a list of attributes. The person's mailbox and advisor, when known, are the remaining elements of the main list.

Within the fictional student file, it is worthwhile to note several characteristics which add complexity to this placement problem:

I.M.A.Person took only the ACT's, while Mickey Mouse took only the SAT's.
Popeye the Sailor took both tests, and those test scores are consistent, while Tweety Bird got extremely different types of scores.
While most transcripts show four years of high school, Tweety Bird's shows only three years and Popeye the Sailor's shows three and a half years. For Tweety Bird, either the school is a three-year high school or the fourth year has not been reported. For Popeye the Sailor, the final semester probably has not yet be reported.

Such variations are common in applications involving expert systems.

Note that the list of attributes is a list of pairs, where each entry (e.g., (ACT 28)) gives the type of information (e.g., ACT first, followed by the data. Such a list of pairs is called an association list, and such lists are a particularly common mechanism for storing data within expert systems. We will see more about processing data within an assocation lists in a later lab.

This example of files illustrates a common circumstance in computing, where data from one application (e.g., the Registrar's file) comes in one format, while data for another application (e.g., the placement program) requires a second format.

Challenge Problem: Write a program which reads a data file in the Registrar's format and produces a corresponding file in the list-oriented format.

This document is available on the World Wide Web as

http://www.math.grin.edu/~walker/courses/153/lab-placement.html

created January 8, 1998
last revised November 1, 1998