PDF pending restoration
ERIC Number: ED279568
Record Type: RIE
Publication Date: 1986-Aug
Reference Count: 0
The Role of Inductive Expert Systems Generators in the Social Science Research Process.
Garson, G. David
Most of what may be construed as political science has always proceeded through induction, which depends heavily on the insight, intuition, and personal brilliance of the particular author. Another approach that is closely associated with the early induction through examples is the expert system. The outcomes of this approach are not measures of the strength of association among variables, but are sets of rules which explain all or perhaps most of the given examples. Expert systems, however, are not a panacea. There is controversy about which algorithms are best and most efficient in automating the generation of expert rule sets. The computer software used in this essay to illustrate social science application of expert systems is based on classification and rule development procedures called "ID 3," developed by J. Ross Quinlan (1983) on the basis of work by Hunt and Stone (1966) on the "Concept of Learning System." It is 1st-Class, Version 3, from Programs in Motion, Inc., which runs on the IBM PC/XT/AT series. The example selected is a reanalysis of data presented by Wayne L. Francis on factors thought to determine the variance in the extent to which state legislators are satisfied with legislative outcomes. Expert systems suggest that it is appropriate to have a long time to consider bills and reject many of them. It is inappropriate to have a short time to consider bills and then accept most. If time is short, it is still inappropriate if one is being inundated with bills from the other chamber. (BZ)
Publication Type: Reports - Research; Speeches/Meeting Papers
Education Level: N/A
Authoring Institution: N/A
Note: Paper presented at the Annual Meeting of the American Political Science Association (Washington, DC, August 28-31, 1986).