Recognizing Textual Entailment Based on WordNet

Textual entailment recognition (RTE) is one of the fundamental problems in many natural language processing applications. This paper proposes a new method for lexical entailment measure which is based on exploiting the information in the WordNet glosses. Further we perform textual entailment recognition based on this method and cast the RTE problem to be a classification problem. The experimental result on RTE challenge data sets indicates that our method is promising.