A cost-reducing question-selection algorithm for propositional knowledge-based systems

Abstract

In many applications of knowledge-based systems, initially given data are often not sufficient to reach a conclusion and more data are needed. A question-selection algorithm is to identify missing information and select proper questions to ask. We present a question-selection algorithm for propositional knowledge-based systems, which aims at asking more relevant and less expensive questions. Comparing to those algorithms currently used in many expert systems, the new algorithm is capable of reaching a conclusion more economically in our computational experiments.