Recognize Textual Entailment by the lexical and semantic matching

Recognizing Textual Entailment (RTE) task is defined as recognizing, given two text fragments, whether the meaning of one text can be inferred from the other. This paper adopts the word overlap technique at first for detection the entailment relation, and then the semantic information is also combined for this task by the use of WordNet which is a structured lexical knowledge base. The part-of-speech and named entity can give useful indication for entailment recognition and we give additional attention to them. The experiments on the RTE test set get the encouraged results and we give the result analysis and we will mine the deeper semantic information in the future work.