Journal
Scientific and technical journal of information technologies, mechanics and optics
UDK004.89
Issue:2 (66)
The paper concerns to the problem of combinatorial complexity of inference in artificial intelligence applications based on production rules. An approach to speeding up the knowledge acquisition is proposed that specifies indexing the facts and operations with sets of indices to prune unnecessary facts. In contrast to the most known algorithms that provide pre-collecting facts in nodes of the search tree, the indices exist apart from the rule sets, that allows disjointing bases of facts and bases of rules logically and physically and to simplify the knowledge base updating. An opportunity of alternation the matching оf rules to set-theoretic operation on tuples of variables is proven.