The Design of an Expert System for Academic Advising at a Two-Year Community College
Date of Award
Doctor of Philosophy (PhD)
Graduate School of Computer and Information Sciences
Steven R. Terrell
George K. Fornshell
This study involves the design of an expert system for academic advisement. Expert systems model the approach that a human expert would take in solving a problem. Expert systems have made important contributions in many areas by applying artificial intelligence technology to a variety of areas of human problem solving. Academic advisement seems to be an appropriate domain for this technology.
Today two-year colleges face challenges offering a sound academic advising program. The colleges must provide faculty advisors who are knowledgeable about academic requirements. These faculty advisors must organize and disseminate complex student information. This information includes whether development classes are needed, if prerequisites to classes have been fulfilled. The advisors must also determine if a class will count toward graduation, and if a class will transfer to a four-year in situation. Faculty advisors are overwhelmed with the sheer volume of information for which they are responsible. Faced with the increasing complexity of institutional policies, procedures and curricula, and articulation agreements with other institutions, faculty advisors struggle to keep up with the constant changes. The advisors feel unprepared with minimal training. Moreover, necessary resources, such as ASSET scores, courses completed, and transfer institution requirements needed for advising are not readily available.
The composition of the student body has changed due to increased numbers of minorities, women, veterans, educationally disadvantaged students, low income students, disabled students, and adult students. All of these students have expressed a need for increased access to academic advising and more individual plans of study.
The changing characteristics of the students, the complexities of our institutions of learning, perceptions of advising, and our current advising system reinforce the need to introduce new technology into academic advising. With all the problems associated with academic advising, especially at the two-year college level, faculty advisors are seeking help with the advising process. Computer assisted advising, using an expert system, offers the promise of a solution to one of the most fundamental advising problems: getting accurate and up-to-date information to students and advisors, and dispensing this information to be the most beneficial to the student. An expert system designed for the advisement of two-year community college students seems to be an appropriate domain for this technology.
This study details the design, development, implementation, and validation of an expert system for the advisement of two-year college students. A rule based prototype expert system was developed, implemented, and then validated using a twostep process. The validation phase utilized four human expert advisors and the expert system for academic advising. Two human experts and the expert system were used in the first step of validation. The records of 30 students to provide the data for the first step. Based on the data in the records each human expert provided academic advisement and completed advising forms for fifteen students. The expert system then provided academic advisement for all 30 students.
The second step utilized the remaining two human experts, as evaluators, and the academic advising forms completed in the first step. The advising forms completed in step one were separated into two sets: one set from the human experts and the second set from the expert system. Fifteen advising forms from each set were selected and placed in a folder along with the matching student records. The remaining fifteen forms from each set were placed in another folder with the matching student records. Each human expert received a folder and were asked to evaluate the courses were selected for the students by the human experts and the academic advisor. The two evaluators independently reviewed, and judged the advisement produced by the two human experts and the academic advising system in step one. The evaluators were instructed to grade the advisement forms on a four-category rating scale: ideal, acceptable, less than acceptable, and unacceptable. The 60 ratings, one set of 30 from each of the reviewing advisors, were then used to determine the extent of difference between the ratings assigned by the advisors. The expert system produced output that was not significantly different from the human experts.
William R. Groover. 1999. The Design of an Expert System for Academic Advising at a Two-Year Community College. Doctoral dissertation. Nova Southeastern University. Retrieved from NSUWorks, Graduate School of Computer and Information Sciences. (553)