Frame Semantics guided network for Abstractive Sentence Summarization

Yong Guan, Shaoru Guo, Ru Li, Xiaoli Li, Hu Zhang

Abstract: Abstractive Text Summarization is an important and practical task, aiming to rephrase the input text into a short version summary, while preserving its same and important semantics. In this paper, we propose a novel Frame Semantics guided network for Abstractive Sentence Summarization (FSum), which is able to learn a better text semantic representation by selecting more relevant Frame semantics from text, and integrating Frame semantic representation with text representation effectively. Extensive experiments demonstrate that our proposed FSum model performs significantly better than existing state-of-the-art techniques on both Gigaword and DUC 2004 benchmark datasets.