Abstract: Many artificial intelligence problems are NP-complete ones. To decrease the needed time of such a problem solving a method of extraction of sub-formulas characterizing the common features of objects under consideration is suggested. This method is based on the offered by the author notion of partial deduction. Repeated application of this procedure allows to form a level description of an object and of classes of objects. A model example of such a level description and the degree of steps number increasing is presented in the paper.

Keywords: Artificial Intelligence, pattern recognition, predicate calculus, level description of a class

Link:

PARTIAL DEDUCTION IN PREDICATE CALCULUS AS A TOOL FOR ARTIFICIAL INTELLIGENCE PROBLEM COMPLEXITY DECREASING

Tatiana M. Kosovskaya

http://www.foibg.com/ijima//vol05/ijima05-03-p05.pdf