Intelligent query mechanism for expert systems

Abstract

The authors describe a set of query techniques which helps to improve the efficiency of the query mechanism of expert systems. They discuss a number of algorithms and formulae for identifying a key feature set and deriving an ideal query sequence. The proposed algorithms aim to cut down the consultation time so as to increase the efficiency of the expert system. The query mechanism can be further enhanced by taking the cost factor of answering each query into consideration. This consideration will affect the value of the features, for the greater the cost required to obtain the answer, the less favorable is the query about that feature. This will lead to more realistic algorithms for finding a KFS (key feature set) and an ideal query sequence. The advantages and disadvantages of a static query mechanism are examined. The solution suggested is to employ a semi-dynamic mechanism. In this way, an appropriate threshold value for the number of patterns which remain undifferentiated has to be determined for the switching from the static to the dynamic mechanism before the suggested mechanism can improve the module.<>