A Multi-View Abstractive Summarization Model Jointly Considering Semantics and Sentiment

Short news summarization is a crucial research hotspot in text summarization. Most work only consider semantic information which can make the summary express the right idea of the original article. However, a good summary should not only deliver the main content, but also its sentiment information. Hence in this paper, we mainly propose a Multi-view abstractive summarization model which can generate summary jointly considering two different views, semantic view and sentiment view. We use encoder-decoder recurrent neural networks for semantic view, and propose two new modules for sentiment view, namely, Sentiment Embedding(SE) and Sentiment Memory(SM). We compare our proposed model with several other summarization models on the Guardian Corpus. The results show that our proposed model performs better than other models. To our best knowledge, it is quite rare and novel to combine large-scale abstractive summarization with sentiment features.

[1]  Hang Li,et al.  A Deep Memory-based Architecture for Sequence-to-Sequence Learning , 2015 .

[2]  Julien Perez,et al.  Gated End-to-End Memory Networks , 2016, EACL.

[3]  Quoc V. Le,et al.  Addressing the Rare Word Problem in Neural Machine Translation , 2014, ACL.

[4]  Zhiyuan Liu,et al.  A C-LSTM Neural Network for Text Classification , 2015, ArXiv.

[5]  Jeffrey Dean,et al.  Distributed Representations of Words and Phrases and their Compositionality , 2013, NIPS.

[6]  Jason Weston,et al.  A Neural Attention Model for Abstractive Sentence Summarization , 2015, EMNLP.

[7]  Jason Weston,et al.  Key-Value Memory Networks for Directly Reading Documents , 2016, EMNLP.

[8]  Bowen Zhou,et al.  Abstractive Text Summarization using Sequence-to-sequence RNNs and Beyond , 2016, CoNLL.

[9]  Yoshua Bengio,et al.  Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation , 2014, EMNLP.

[10]  Udo Kruschwitz,et al.  MultiLing 2015: Multilingual Summarization of Single and Multi-Documents, On-line Fora, and Call-center Conversations , 2015, SIGDIAL Conference.

[11]  Chin-Yew Lin,et al.  ROUGE: A Package for Automatic Evaluation of Summaries , 2004, ACL 2004.

[12]  Yoshua Bengio,et al.  Neural Machine Translation by Jointly Learning to Align and Translate , 2014, ICLR.

[13]  Wei Heng,et al.  CIST System Report for ACL MultiLing 2013 – Track 1: Multilingual Multi-document Summarization , 2013 .

[14]  Kuldip K. Paliwal,et al.  Bidirectional recurrent neural networks , 1997, IEEE Trans. Signal Process..

[15]  Matthew D. Zeiler.  ADADELTA: An Adaptive Learning Rate Method , 2012, ArXiv.