Search Strategies in the State Space of Knowledge Bases

Abstract

Search strategies in the state space of knowledge bases in intellectual systems are considered. The directions of search from the initial data of the task to the goal and in reverse direction are shown. Rules and admissible moves leading to the goal in certain conditions of their application, when they become new search goals or subgoals, are analyzed. The problem-solving module is used as a search strategy for both data-driven and goal-driven search. It is shown that the choice of goal depends on the structure of the problem to be solved. The method of embedding the thesaurus into a probabilistic model for optimizing information retrieval is described.