A hybrid textual entailment system using lexical and syntactic features

A two-way textual entailment (TE) recognition system that uses lexical and syntactic features has been described in this paper. The hybrid TE system is based on the Support Vector Machine that uses twenty three features for lexical similarity and the output tag from a rule based syntactic two-way TE system as another feature. The important lexical features that are used in the present system are: WordNet based unigram match, bigram match, longest common subsequence, skip-gram, stemming, named entity matching and lexical distance. In the syntactic TE system, the important features used are: subject-subject comparison, subject-verb comparison, object-verb comparison and cross subject-verb comparison. The hybrid system has been developed using the collection of RTE-2 test annotated set, RTE-3 development set and RTE-3 test gold set that includes 2400 text-hypothesis pairs. Evaluation scores obtained on the RTE-4 test set (includes 1000 text-hypothesis pairs) show 55.30% precision and 58.40% recall for YES decisions and 55.93% precision and 52.80% recall for NO decisions.