|
|
|
| A Deep-Learning Model Based on Attention Mechanism for Chinese Comparative Relation Detection |
| Zhu Maoran1, Wang Yilei1, Gao Song2, Wang Hongwei1, Zheng Lijuan3 |
1.School of Economics and Management, Tongji University, Shanghai 200092
2.China Information Technology Security Evaluation Center, Beijing 100085
3.School of Business, Liaocheng University, Liaocheng 252000 |
|
|
|
|
Abstract There are a massive number of comparative opinions in online reviews, containing users’ assessments of their experience with different products or services. Businesses can gain insight into their market competitiveness by identifying useful user-generated comparison information from among a mass of low-quality comments. We were thereby motivated to study comparative sentence recognition in Chinese comments. Instead of using the pattern recognition method as past studies did, we addressed the recognition task through a hierarchical multi-attention network based on deep learning. Our model outperforms both the traditional classification model and the deep-learning-based text classification model in term of accuracy, with the F1-score reaching 81%. The proposed hierarchical multi-attention network model is end-to-end, thus avoiding the design of a large number of artificial features, and greatly reducing human involvement for comparative comment recognition.
|
|
Received: 16 December 2018
|
|
|
|
1 QaziA, SyedK B S, RajR G, et al. A concept-level approach to the analysis of online review helpfulness[J]. Computers in Human Behavior, 2016, 58: 75-81.
2 GaoS, TangO, WangH, et al. Identifying competitors through comparative relation mining of online reviews in the restaurant industry[J]. International Journal of Hospitality Management, 2018, 71: 19-32.
3 VarathanK D, GiachanouA, CrestaniF. Comparative opinion mining: a review[J]. Journal of the Association for Information Science and Technology, 2017, 68(4): 811-829.
4 JindalN, LiuB. Identifying comparative sentences in text documents[C]// Proceedings of the 29th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval. New York: ACM Press, 2006: 244-251.
5 吕叔湘. 吕叔湘文集: 中国文法要略[M]. 北京: 商务印书馆, 1990.
6 刘焱. “比”字句对比较项选择的语义认知基础[J]. 上海财经大学学报, 2004, 6(5): 76-80.
7 ParkD H, BlakeC. Identifying comparative claim sentences in full-text scientific articles[C]// Proceedings of the Workshop on Detecting Structure in Scholarly Discourse. Stroudsburg: Association for Computational Linguistics, 2012: 1-9.
8 黄小江, 万小军, 杨建武, 等. 汉语比较句识别研究[J]. 中文信息学报, 2008, 22(5): 30-38.
9 黄高辉, 姚天昉, 刘全升. CRF算法的汉语比较句识别和关系抽取[J]. 计算机应用研究, 2010, 27(6): 2061-2064.
10 李建军, 何中市. 比较句与比较关系识别研究及其应用[D]. 重庆: 重庆大学, 2011.
11 张辰, 冯冲, 刘全超. 基于多特征融合的中文比较句识别算法[J]. 中文信息学报, 2013, 27(6): 110-117.
12 WangH W, GaoS, YinP, et al. Competitiveness analysis through comparative relation mining: Evidence from restaurantsonline reviews[J]. Industrial Management & Data Systems, 2017, 117(4): 672-687.
13 王巍, 赵铁军, 徐冰, 等. 中文比较句的自动识别[J]. 智能计算机与应用, 2015(5): 1-3.
14 KimY. Convolutional neural networks for sentence classification[C]// Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing. Stroudsburg: Association for Computational Linguistics, 2014: 1746-1751.
15 ConneauA, SchwenkH, BarraultL, et al. Very deep convolutional networks for natural language processing[OL]. https://arxiv.org/pdf/1606.01781.pdf.
16 SocherR, LinC C Y, NgA Y, et al. Parsing natural scenes and natural language with recursive neural networks[C]// Proceedings of the 28th International Conference on Machine Learning. Madison: Omnipress, 2011: 129-136.
17 TaiK S, SocherR, ManningC D. Improved semantic representations from tree-structured long short-term memory networks[OL]. https://arxiv.org/pdf/1503.00075.pdf.
18 ZhouC T, SunC L, LiuZ Y, et al. A C-LSTM neural network for text classification[OL]. https://arxiv.org/pdf/1511.08630.pdf.
19 YangZ C, YangD Y, DyerC, et al. Hierarchical attention networks for document classification[C]// Proceedings of the 2016 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. Stroudsburg: Association for Computational Linguistics, 2016: 1480-1489.
20 LinZ H, FengM W, dos SantosC N, et al. A structured self-attentive sentence embedding[OL]. https://arxiv.org/pdf/1703.03130.pdf.
21 PappasN, Popescu-BelisA. Multilingual hierarchical attention networks for document classification[C]// Proceedings of the 8th International Joint Conference on Natural Language Processing. Stroudsburg: Association for Computational Linguistics, 2017: 1015-1025.
22 MikolovT, ChenK, CorradoG, et al. Efficient estimation of word representations in vector space[OL]. https://arxiv.org/pdf/1301.3781.pdf.
23 来斯惟. 基于神经网络的词和文档语义向量表示方法研究[D]. 北京: 中国科学院大学, 2016.
24 MelamudO, McCloskyD, PatwardhanS, et al. The role of context types and dimensionality in learning word embeddings[C]// Proceedings of the Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. Stroudsburg: Association for Computational Linguistics, 2016: 1030-1040.
25 BahdanauD, ChoK, BengioY. Neural machine translation by jointly learning to align and translate[OL]. https://arxiv.org/pdf/1409.0473v7.pdf. |
|
|
|
