The development and validation of an expert system for academic advisement /
| Main Author: | |
|---|---|
| Other Authors: | , , |
| Format: | Thesis Book |
| Language: | English |
| Published: |
1991.
|
| Subjects: | |
| Online Access: | ProQuest, Abstract Link to OAKTrust copy |
| Abstract: | This study involved the development and validation of an expert system for academic advisement. Academic advisement has been recognized as an important function of higher education. However, there is a widespread desire to improve the advisement services provided by many colleges and universities. At the beginning of the 1990s, expert systems had made important contributions in many domains by applying artificial intelligence to a variety of areas of human problem solving. Although few expert advisement systems existed at the time of this study, academic advisement seemed to be an appropriate domain for this technology. An expert system called the Academic Advisor (AA) was developed using methodology advocated in the literature. Problem identification, conceptualization, implementation, and validation were major steps in the development of this system. The validation component of this study included a two-phase design for assessing the accuracy of AA's output. Six human advisors and the AA system provided advice for 10 test case students. The total body of advice provided by all experts consisted of 988 statements of advice. The first phase of validation was a comparison of AA's output with that of the human experts. The findings indicated that AA's advice matched the advice provided by the human advisors 266 times. Compared with the results of the six human advisors, the expert system produced the third highest number of corroborated statements. In the second validation phase, three "blinded" evaluators rated (as ideal, acceptable, less than acceptable, or unacceptable) the advice produced by the six human advisors and AA. Although the proportion of agreement between the evaluators' ratings was low (.46), an average of 67 percent of AA's advice was judged to be acceptable to them. There was no significant statistical difference at the .05 level of probability between the performance of the most highly rated human advisors and the expert system. Conclusions from the study indicated that an expert system would be a feasible means to support academic advisement in a two-year college setting. |
|---|---|
| Item Description: | Typescript (photocopy). Vita. "Major subject: Industrial Education." |
| Physical Description: | ix, 258 leaves : illustrations ; 29 cm |
| Bibliography: | Includes bibliographical references. |