Issue No. 02 - April (1993 vol. 5)
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/69.219743
<p>An approach for representing inference control is presented. It is proposed that the representation of inference control should consist of two levels: planning level which realizes problem solving strategies, and a performing level, which represents inference tactics. Based on this approach, the representation system hypothesis-based associative representation (HAR) has been developed to realize the functional architecture for knowledge-based systems. Because users are allowed to organize hypothesis-based associative networks that perform the problem solving strategies with different features, HAR becomes not only a tool for building knowledge-based systems, but also an environment for exploring AI techniques. For example, by comparing three strategies of block-world action planning, it is found that the least commitment strategy is the most efficient.</p>
inference control; planning level; problem solving strategies; performing level; inference tactics; representation system hypothesis-based associative representation; HAR; functional architecture; knowledge-based systems; hypothesis-based associative networks; AI techniques; block-world action planning; least commitment strategy; inference mechanisms; knowledge based systems; knowledge representation
G. Ji, "Representing Inference Control by Hypothesis-Based Association," in IEEE Transactions on Knowledge & Data Engineering, vol. 5, no. , pp. 363-367, 1993.