A Text Summarization Approach under the Influence of Textual Entailment

Abstract

This paper presents how text summarization can be influenced by textual entailment. We show that if we use textual entailment recognition together with text summarization approach, we achieve good results for final summaries, obtaining an improvement of 6.78% with respect to the summarization approach only. We also compare the performance of this combined approach to two baselines (the one provided in DUC 2002 and ours based on word-frequency technique) and we discuss the preliminary results obtained in order to infer conclusions that can be useful for future research.