Fuzzy Logic based Enhanced Word Sense Identification through Relation Construction

Abstract

The paper explores the use of a fuzzy based reasoning algorithm for identifying the exact sense of a word in a Question Answering system. The intelligent formulation encapsulates the process of learning through the comparison of related words that occur in a sentence. The procedure establishes a fuzzy set for the target word with one polysemous word in the sentence using the Word Net corpus and attempts to derive its exact sense based on the membership degree. The association between the observed word pair and the one available in the corpus quantifies a degree of membership of t_w for the set of words that are assorted as senses for t_w. It promotes the sequence of steps in multiple contexts to create a directive through the fuzzy relation matrix. The results obtained exhibit its superior capability over the supervised methods and project its utility in the real world applications.