Neural sentence fusion for diversity driven abstractive multi-document summarization

摘要

The lack of multi-document based models and the inaccuracy in representing multiple long documents into a fixed size vector inspired us to solve abstractive multi-document summarization. Also, there is lack of good multi-document based human-authored datasets to train any encoder-decoder models. To overcome this, we have designed complementary models for two different tasks such as sentence clustering and neural sentence fusion. In this work, we minimize the risk of producing incorrect fact by encoding a related set of sentences as an input to the encoder. We have applied our complementary models to implement a full abstractive multi-document summarization system which simultaneously considers importance, coverage, and diversity under a desired length limit. We conduct extensive experiments for all the proposed models which bring significant improvements over the state-of-the-art methods across different evaluation metrics. (C) 2019 Elsevier Ltd. All rights reserved.