A Sentence Summarizer using Recurrent Neural Network and Attention-Based Encoder

DOI

10.2991/amms-17.2017.54How to use a DOI?

Keywords

abstractive summarization; recurrent neural network; auto-encoder; nature language understanding; artificial intelligence

Abstract

For automatically summarizing sentences of nature languages, some cutting-age methods have been proposed since a decade ago. In this paper, an advanced model of abstractive sentence summarization is proposed by composing a recurrent neural network (RNN) and an attention-based encoder. The proposed model is an improvement version of Rush-Chopra-Weston's neural attention model, and main differences between the proposed model and the conventional one is that: 1) the novel model utilizes two RNNs instead of the feed-forward neural networks; 2) the length of summarized sentence (the output of these models) is variable (which is fixed in the conventional case). Experiments showed the effectiveness of the proposed sentence summarizer and these results suggest that it is possible to abstract long articles into shorten words.

Copyright

© 2017, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Takashi Kuremoto
AU  - Takuji Tsuruda
AU  - Shingo Mabu
AU  - Masanao Obayashi
PY  - 2017/11
DA  - 2017/11
TI  - A Sentence Summarizer using Recurrent Neural Network and Attention-Based Encoder
BT  - Proceedings of the 2017 International Conference on Applied Mathematics, Modeling and Simulation (AMMS 2017)
PB  - Atlantis Press
SP  - 245
EP  - 248
SN  - 1951-6851
UR  - https://doi.org/10.2991/amms-17.2017.54
DO  - 10.2991/amms-17.2017.54
ID  - Kuremoto2017/11
ER  -