Fine-Grained Opinion Extraction from Chinese Car Reviews with an Integrated Strategy

doi:10.1007/s12204-018-1961-6

Journal of Shanghai Jiao Tong University (Science) ›› 2018, Vol. 23 ›› Issue (5): 620-626.doi: 10.1007/s12204-018-1961-6

Fine-Grained Opinion Extraction from Chinese Car Reviews with an Integrated Strategy

WANG Yinglin (王英林), WANG Ming (王明)

(School of Information Management and Engineering, Shanghai University of Finance and Economics, Shanghai 200433, China)

出版日期:2018-10-01 发布日期:2018-10-07
通讯作者: WANG Yinglin (王英林) E-mail: wang.yinglin@shufe.edu.cn

Fine-Grained Opinion Extraction from Chinese Car Reviews with an Integrated Strategy

WANG Yinglin (王英林), WANG Ming (王明)

(School of Information Management and Engineering, Shanghai University of Finance and Economics, Shanghai 200433, China)

Online:2018-10-01 Published:2018-10-07
Contact: WANG Yinglin (王英林) E-mail: wang.yinglin@shufe.edu.cn

摘要/Abstract

摘要： With rapid development of E-commerce, a large amount of data including reviews about different types of products can be accessed within short time. On top of this, opinion mining is becoming increasingly effective to extract valuable information for product design, improvement and brand marketing, especially with fine-grained opinion mining. However, limited by the unstructured and causal expression of opinions, one cannot extract valuable information conveniently. In this paper, we propose an integrated strategy to automatically extract feature-based information, with which one can easily acquire detailed opinion about certain products. For adaptation to the reviews’ characteristics, our strategy is made up of a multi-label classification (MLC) for reviews, a binary classification (BC) for sentences and a sentence-level sequence labelling with a deep learning method. During experiment, our approach achieves 82% accuracy in the final sequence labelling task under the setting of a 20-fold cross validation. In addition, the strategy can be expediently employed in other reviews as long as there is an according amount of labelled data for startup.

关键词: opinion extraction, multi-label classification (MLC), binary classification (BC), sequence labelling, recurrent neural network (RNN)

Abstract: With rapid development of E-commerce, a large amount of data including reviews about different types of products can be accessed within short time. On top of this, opinion mining is becoming increasingly effective to extract valuable information for product design, improvement and brand marketing, especially with fine-grained opinion mining. However, limited by the unstructured and causal expression of opinions, one cannot extract valuable information conveniently. In this paper, we propose an integrated strategy to automatically extract feature-based information, with which one can easily acquire detailed opinion about certain products. For adaptation to the reviews’ characteristics, our strategy is made up of a multi-label classification (MLC) for reviews, a binary classification (BC) for sentences and a sentence-level sequence labelling with a deep learning method. During experiment, our approach achieves 82% accuracy in the final sequence labelling task under the setting of a 20-fold cross validation. In addition, the strategy can be expediently employed in other reviews as long as there is an according amount of labelled data for startup.

Key words: opinion extraction, multi-label classification (MLC), binary classification (BC), sequence labelling, recurrent neural network (RNN)

中图分类号:

TP 391.1

WANG Yinglin (王英林), WANG Ming (王明). Fine-Grained Opinion Extraction from Chinese Car Reviews with an Integrated Strategy[J]. Journal of Shanghai Jiao Tong University (Science), 2018, 23(5): 620-626.

参考文献 18

[1]	SUN S L, LUO C, CHEN J Y. A review of naturallanguage processing techniques for opinion mining systems[J]. Information Fusion, 2016, 36: 10-25.
[2]	YANG B H, CARDIE C. Joint inference for finegrainedopinion extraction [C]//Proceedings of the 51stAnnual Meeting of the Association for ComputationalLinguistics. Sofia, Bulgaria: Association for ComputationalLinguistics, 2013: 1640-1649.
[3]	WIEBE J, WILSON T, CARDIE C. Annotatingexpressions of opinions and emotions in language[J].Language Resources and Evaluation, 2005, 39(2/3):165-210.
[4]	HUANG Z H, XU W, YU K. Bidirectional LSTMCRFmodels for sequence tagging [EB/OL]. (2015-08-09) [2018-03-05]. https://arxiv.org/abs/1508.01991v1.
[5]	SHEN X P, BOUTELL M, LUO J B, et al. Multilabelmachine learning and its application to semanticscene classification [C]//Proceedings of SPIE-IS & TElectronic Imaging. [s.l.]: SPIE, 2004: 18-22.
[6]	ZHANG J, LI D Y,WANG S G.Multiple performancesidentification for car review texts based on multi-labellearning [J]. Computer Engineering and Science, 2016,38(1): 188-194 (in Chinese).
[7]	ZHANG M L, ZHOU Z H. A review on multi-labellearning algorithms [J]. IEEE Transactions on Knowledgeand Data Engineering, 2014, 26(8): 1819-1837.
[8]	LAFFERTY J, MCCALLUM A, PEREIRA F. Conditionalrandom fields: Probabilistic models for segmentingand labeling sequence data [C]//EighteenthInternational Conference on Machine Learning. SanFrancisco, USA: Morgan Kaufmann Publishers, 2001:282-289.
[9]	PONTIKI M, GALANIS D, PAVLOPOULOS J, et al.SemEval-2014 task 4: Aspect based sentiment analysis[C]//Proceedings of the 8th International Workshop onSemantic Evaluation. Dublin, Ireland: [s.n.], 2014: 27-35.
[10]	˙IRSOY O, CARDIE C. Opinion mining with deep recurrentneural networks [C]//Proceedings of the 2014Conference on Empirical Methods in Natural LanguageProcessing (EMNLP). Doha, Qatar: Association forComputational Linguistics, 2014: 720-728.
[11]	LIU P, JOTY S, MENG H. Fine-grained opinion miningwith recurrent neural networks and word embeddings[C]//Proceedings of the 2015 Conference on EmpiricalMethods in Natural Language Processing. Lisbon,Portugal: Association for Computational Linguistics,2015: 1433-1443.
[12]	BRECK E, CHOI Y, CARDIE C. Identifying expressionsof opinion in context [C]//Proceedings of the20th International Joint Conference on Artifical Intelligence.Hyderabad, India: ACM, 2007: 2683-2688.
[13]	WEI J, HUNG H H, ROHINI K. A novel lexicalizedHMM-based learning framework for web opinion mining[C]//Proceedings of the 26th International Conferenceon Machine Learning. Montreal, Canada: [s.n.],2009: 465-472.
[14]	SAMANEH M, MARTIN E. On the design ofLDA models for aspect-based opinion mining[C]//Proceedings of CIKM. Maui, USA: ACM, 2012;803-812.
[15]	MADJAROV G, KOCEV D, GJORGJEVIKJ D, etal. An extensive experimental comparison of methodsfor multi-label learning [J]. Pattern Recognition, 2012,45(9): 3084-3104.
[16]	HOCHREITER S, SCHMIDHUBER J. Long shorttermmemory [J]. Neural Computation, 1997, 9(8):1735-1780.
[17]	GRAVES A, SCHMIDHUBER J. Framewise phonemeclassification with bidirectional LSTM and other neuralnetwork architectures [J]. Neural Networks, 2005,18(5): 602-610.
[18]	ZHANG M L, PE?NA J M, ROBLES V. Feature selectionfor multi-label naive bayes classification [J]. InformationSciences, 2009, 179(19): 3218-3229.

Fine-Grained Opinion Extraction from Chinese Car Reviews with an Integrated Strategy

Fine-Grained Opinion Extraction from Chinese Car Reviews with an Integrated Strategy

可视化

摘要/Abstract

引用本文

使用本文

参考文献 18

相关文章 1

编辑推荐

Metrics

本文评价