On the knowledge-based pattern recognition using syntactic approach

摘要

This paper presents a method for building the knowledge-based pattern recognition system using syntactic approach. Two important problems are discussed: the knowledge representation by means of grammar production and the search strategy. A production of context-free grammar [A → β] can represent knowledge [IF β THEN A]. This conclusion is just contrary to the conventional viewpoint. For the heuristic search a depth-first parsing algorithm—the ED algorithm—is presented. On this basis the author has proposed a new type of knowledge-based pattern recognition system, in which the attributed grammar is used to represent knowledge and the Earley algorithm or the ED algorithm is used to search the conclusion. Two examples of application are given.