AMBIGUITY RESOLUTION USING LEXICAL ASSOCIATION

Lexical information has been shown to be crucial for decisions on ambiguities. Many statistical parsers is based on probabilities of this dependencies. Our system tried to conjoin lexical information to the best first parsing method and to show that every nodes can be determined using GAT(global association table) ,which is a new data structure to manage the lexical associations. In Korean, the structual ambiguity and the grammatical case ambiguity influence the accuracy of the parser. Lexical information between pairs of words is computed by co-occurrence data extracted from the corpus and to be extended to the conceptual association with thesaurus ,which it attempts to reduce parameter space.