AI helps you reading Science

AI generates interpretation videos

AI extracts and analyses the key points of the paper to generate videos automatically


pub
Go Generating

AI Traceability

AI parses the academic lineage of this thesis


Master Reading Tree
Generate MRT

AI Insight

AI extracts a summary of this paper


Weibo:
The novelties of our work included: 1) we defined the phrase dependency parsing and proposed an approach to construct the phrase dependency trees; 2) we proposed a new tree kernel function to model the phrase dependency trees

Phrase dependency parsing for opinion mining

EMNLP, pp.1533-1541, (2009)

Cited by: 425|Views143
EI WOS SCOPUS
Full Text
Bibtex
Weibo

Abstract

In this paper, we present a novel approach for mining opinions from product reviews, where it converts opinion mining task to identify product features, expressions of opinions and relations between them. By taking advantage of the observation that a lot of product features are phrases, a concept of phrase dependency parsing is introduced...More

Code:

Data:

Introduction
  • As millions of users contribute rich information to the Internet everyday, an enormous number of product reviews are freely written in blog pages, Web forums and other consumer-generated mediums (CGMs).
  • This vast richness of content becomes increasingly important information source for collecting and tracking customer opinions.
  • Comparing with the former one, opinion mining usually produces richer information
Highlights
  • As millions of users contribute rich information to the Internet everyday, an enormous number of product reviews are freely written in blog pages, Web forums and other consumer-generated mediums (CGMs)
  • Results of extracting product features and opinion expressions are shown in Table 2
  • The candidate product features are extracted by the method described in Section 2.2, whose result is in the first row. 6760 of 24414 candidate product features remained after the filtering, which means we cut 72% of irrelevant candidates with a cost of 14.5%(1-85.5%) loss in true answers
  • We focused on extracting relations between product features and opinion expressions
  • The novelties of our work included: 1) we defined the phrase dependency parsing and proposed an approach to construct the phrase dependency trees; 2) we proposed a new tree kernel function to model the phrase dependency trees
  • Experimental results show that our approach improved the performances of the mining task
Methods
  • P R F Adjacent SVM-1 SVM-2 OERight PFRight MP3 Player P RF.
  • Digital Camera PR F DVD Player Diaper PR F PRF.
  • Adjacent SVM-1 SVM-2 SVM-WTree SVM-PTree PR DVD Player PR F MP3 Player
Results
  • Results of extracting product features and opinion expressions are shown in Table 2.
  • Similar to the product feature extraction, the precision of extracting opinion expression is relatively low, while the recall is 75.2%.
  • A further inspection into the result of first 3 domains, the authors can conclude that: 1) Tree kernels(SVM-WTree and SVM-PTree) are better than Adjacent, SVM-1 and SVM-2 in all domains.
  • It proofs that the dependency tree is important in the opinion relation extraction.
  • The authors believe the main reason is that phrase dependency tree provides a more succinct tree structure, and the separative treatment of local dependencies and global dependencies in kernel computation can improve the performance of relation extraction
Conclusion
  • The authors described the work on mining opinions from unstructured documents.
  • The authors focused on extracting relations between product features and opinion expressions.
  • The novelties of the work included: 1) the authors defined the phrase dependency parsing and proposed an approach to construct the phrase dependency trees; 2) the authors proposed a new tree kernel function to model the phrase dependency trees.
  • Experimental results show that the approach improved the performances of the mining task
Tables
  • Table1: Statistics for the annotated corpus m(Ti, Tj) =
  • Table2: Results for extracting product features and opinion expressions
  • Table3: Features used in SVM-1: o denotes an
  • Table4: Features used in SVM-PTree
  • Table5: Results of different methods
  • Table6: Results for total performance with cross domain training data
Download tables as Excel
Related work
  • Opinion mining has recently received considerable attention. Amount of works have been done on sentimental classification in different levels (Zhang et al, 2009; Somasundaran et al, 2008; Pang et al, 2002; Dave et al, 2003; Kim and Hovy, 2004; Takamura et al, 2005). While we focus on extracting product features, opinion expressions and mining relations in this paper.

    Kobayashi et al (2007) presented their work on extracting opinion units including: opinion holder, subject, aspect and evaluation. Subject and aspect belong to product features, while evaluation is the opinion expression in our work. They converted the task to two kinds of relation extraction tasks and proposed a machine learning-based method which combines contextual clues and statistical clues. Their experimental results showed that the model using contextual clues improved the performance. However since the contextual information in a domain is specific, the model got by their approach can not easily converted to other domains.
Funding
  • This work was (partially) funded by Chinese NSF 60673038, Doctoral Fund of Ministry of Education of China 200802460066, and Shanghai Science and Technology Development Funds 08511500302
Reference
  • Eric Breck, Yejin Choi, and Claire Cardie. 2007. Identifying expressions of opinion in context. In Proceedings of IJCAI-2007.
    Google ScholarLocate open access versionFindings
  • Yejin Choi, Claire Cardie, Ellen Riloff, and Siddharth Patwardhan. 2005. Identifying sources of opinions with conditional random fields and extraction patterns. In Proceedings of HLT/EMNLP.
    Google ScholarLocate open access versionFindings
  • Yejin Choi, Eric Breck, and Claire Cardie. 2006. Joint extraction of entities and relations for opinion recognition. In Proceedings EMNLP.
    Google ScholarLocate open access versionFindings
  • Aron Culotta and Jeffrey Sorensen. 200Dependency tree kernels for relation extraction. In In Proceedings of ACL 2004.
    Google ScholarLocate open access versionFindings
  • Kushal Dave, Steve Lawrence, and David M. Pennock. 2003. Mining the peanut gallery: opinion extraction and semantic classification of product reviews. In Proceedings of WWW 2003.
    Google ScholarLocate open access versionFindings
  • Minqing Hu and Bing Liu. 2004. Mining and summarizing customer reviews. In Proceedings of the ACM SIGKDD 2004.
    Google ScholarLocate open access versionFindings
  • Nitin Jindal and Bing Liu. 2008. Opinion spam and analysis. In Proceedings of WSDM ’08.
    Google ScholarLocate open access versionFindings
  • Thorsten Joachims, Nello Cristianini, and John ShaweTaylor. 2001. Composite kernels for hypertext categorisation. In Proceedings of ICML ’01.
    Google ScholarLocate open access versionFindings
  • Soo-Min Kim and Eduard Hovy. 2004. Determining the sentiment of opinions. In Proceedings of Coling 2004. COLING.
    Google ScholarLocate open access versionFindings
  • Dan Klein and Christopher D. Manning. 2002. Fast exact inference with a factored model for natural language parsing. In In Advances in Neural Information Processing Systems.
    Google ScholarLocate open access versionFindings
  • Nozomi Kobayashi, Kentaro Inui, and Yuji Matsumoto. 2007. Extracting aspect-evaluation and aspect-of relations in opinion mining. In Proceedings of EMNLP-CoNLL 2007.
    Google ScholarLocate open access versionFindings
  • Bo Pang, Lillian Lee, and Shivakumar Vaithyanathan. 2002. Thumbs up? Sentiment classification using machine learning techniques. In Proc. of EMNLP 2002.
    Google ScholarLocate open access versionFindings
  • Ana-Maria Popescu and Oren Etzioni. 2005. Extracting product features and opinions from reviews. In Proceedings of HLT/EMNLP.
    Google ScholarLocate open access versionFindings
  • E. Riloff and W. Phillips. 2004. An introduction to the sundance and autoslog systems. In University of Utah School of Computing Technical Report UUCS04-015.
    Google ScholarFindings
  • Beatrice Santorini and Anthony Kroch. 2007.
    Google ScholarFindings
  • http://www.ling.upenn.edu/
    Findings
  • Swapna Somasundaran, Janyce Wiebe, and Josef Ruppenhofer. 2008. Discourse level opinion interpretation. In Proceedings of COLING 2008.
    Google ScholarLocate open access versionFindings
  • Hiroya Takamura, Takashi Inui, and Manabu Okumura. 2005. Extracting semantic orientations of words using spin model. In Proceedings of ACL’05.
    Google ScholarLocate open access versionFindings
  • L. Tesniere. 1959. Elements de syntaxe structurale. Editions Klincksieck.
    Google ScholarLocate open access versionFindings
  • Janyce Wiebe, Theresa Wilson, and Claire Cardie. 2005. Annotating expressions of opinions and emotions in language. Language Resources and Evaluation, 39(2/3).
    Google ScholarLocate open access versionFindings
  • Theresa Wilson, Paul Hoffmann, Swapna Somasundaran, Jason Kessler, Janyce Wiebe, Yejin Choi, Claire Cardie, Ellen Riloff, and Siddharth Patwardhan. 2005a. Opinionfinder: A system for subjectivity analysis. In Demonstration Description in Conference on Empirical Methods in Natural Language Processing.
    Google ScholarLocate open access versionFindings
  • Theresa Wilson, Janyce Wiebe, and Paul Hoffmann. 2005b. Recognizing contextual polarity in phraselevel sentiment analysis. In Proceedings of HLTEMNLP.
    Google ScholarLocate open access versionFindings
  • Fei Xia and Martha Palmer. 2001. Converting dependency structures to phrase structures. In HLT ’01: Proceedings of the first international conference on Human language technology research.
    Google ScholarLocate open access versionFindings
  • Qi Zhang, Yuanbin Wu, Tao Li, Mitsunori Ogihara, Joseph Johnson, and Xuanjing Huang. 2009. Mining product reviews based on shallow dependency parsing. In Proceedings of SIGIR 2009.
    Google ScholarLocate open access versionFindings
Your rating :
0

 

Tags
Comments
数据免责声明
页面数据均来自互联网公开来源、合作出版商和通过AI技术自动分析结果,我们不对页面数据的有效性、准确性、正确性、可靠性、完整性和及时性做出任何承诺和保证。若有疑问,可以通过电子邮件方式联系我们:report@aminer.cn
小科