Abstract:Abstractive summarization is a important task in natural language processing field. In previous work, the sequence-to-sequence based models are widely used for abstractive summarization task. However, most of the current abstractive summarization models still suffer from two problems. One is that it is difficult for these models to learn an accurate source contextual representation from the redundancy and noisy source text at each decoding step. Another is the information loss problem, which is ignored in previous work. The inability of these models to effectively exploit previously generated words led to this problem. In order to address these two problems, in this paper, we propose a novel keyword and generated word attention model. Specifically, the proposed model first employs the hidden state of decoder to capture relevant keywords and previously generated words contextual at each time step. The model then utilizes obtained keywords and generated words contextual to create keywords-aware and generated words-aware source contextual, respectively. The keywords contextual contributes to learn an accurate source contextual representation, and the generated words contextual can alleviate the information loss problem. Experimental results on a popular Chinese social media dataset demonstrate that the proposed model outperforms baselines and achieves the state-of-the-art performance.

Keyphrase Generation With Word Attention

Word Attention for Sequence to Sequence Text Understanding.

Keyphrase Generation with Cross-Document Attention

Neural Keyphrase Generation: Analysis and Evaluation

Abstractive Summarization with Keyword and Generated Word Attention.

Deep Keyphrase Generation

Keyphrase Enhanced Diverse Beam Search: A Content-Introducing Approach To Neural Text Generation

WORD EMBEDDING ATTENTION NETWORK: GENERATING WORDS BY QUERYING DISTRIBUTED WORD REPRESENTATIONS FOR PARAPHRASE GENERATION

Query and Output: Generating Words by Querying Distributed Word Representations for Paraphrase Generation

WordTopic-MultiRank: A New Method for Automatic Keyphrase Extraction.

Title-Guided Encoding for Keyphrase Generation

Keyphrase Generation for Scientific Articles using GANs

A Novel Keyword Generation Model Based on Topic-Aware and Title-Guide

Keyphrase Guided Beam Search for Neural Abstractive Text Summarization.

Heterogeneous Graph Neural Networks for Keyphrase Generation

Keyphrase Generation Beyond the Boundaries of Title and Abstract

SGG: Learning to Select, Guide, and Generate for Keyphrase Generation

Keyphrase Generation: A Text Summarization Struggle

Keyphrase Generation: A Multi-Aspect Survey

Unleashing the Potential of Attention Model for News Headline Generation

Automatic Keyphrase Extraction Via Topic Decomposition.