1 / 11

Fully Automatic Lexicon Expansion for Domain-oriented Sentiment Analysis

Proceedings of the 2006 Conference on Empirical Methods in Natural Language Processing (EMNLP 2006) , pages 355–363. Fully Automatic Lexicon Expansion for Domain-oriented Sentiment Analysis. Hiroshi Kanayama Tetsuya Nasukawa Tokyo Research Laboratory, IBM Japan, Ltd. Abstract.

jewel
Download Presentation

Fully Automatic Lexicon Expansion for Domain-oriented Sentiment Analysis

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Proceedings of the 2006 Conference on Empirical Methods in Natural Language Processing (EMNLP 2006), pages 355–363. Fully Automatic Lexicon Expansionfor Domain-oriented Sentiment Analysis Hiroshi Kanayama Tetsuya Nasukawa Tokyo Research Laboratory, IBM Japan, Ltd.

  2. Abstract • an unsupervisedlexicon building method for the detectionof polar clauses • convey positiveor negative aspects in a specificdomain • lexical entries acquired are called polar atoms • use context coherency, the tendency for same polarities to appear successively in contexts • the overall density and precision (94% in this work) of coherency in the corpus • the statistical estimation picks up appropriate polar atoms among candidates, without any manual tuning of the threshold values

  3. Introduction • Sentiment Analysis (SA) is a task to recognize writers’ feelings as expressed in positive or negative comments, by analyzing unreadably large numbers of documents • Japanese sentence • in the digital camera domainSubjective (SA通常針對主觀句)Kono kamera-ha subarashii-to omou.‘I think this camera is splendid. [+] splendid(camera)Objective (客觀句在特定domain也有用處)Kontorasuto-ga kukkiri-suru.‘The contrast is sharp.’Atarashii kishu-ha zuumu-mo tsuite-iru.‘The new model has a zoom lens, too.’

  4. Methodology of Clause-level SA(syntactic; 先根據日文文法) • sentence delimitation • proposition detection • Polarity assignment このカメラは素晴らしいと思う。 I think this camera is splendid このカメラは美しくない このカメラは美しいとよい 103 domain independent patterns 22 conjunctive patterns

  5. Polar Atoms • a minimum syntactic structure specifying polarity in a predicative expression • [+] kukkiri-suru • ‘to be sharp’ • [+] tsuku <- zuumu-ga • ‘to have <-zoom lens-NOM’ • 3,275個

  6. Context Coherency • 分Inter和Intra • ‘I think this camera is splendid.’‘It’s light and has a zoom lens.’‘Though the LCD is small, I’m satisfied.’ 雖然…‘But, the price is a little high.’但是… • Inter若遇到這些字

  7. context coherency • an assumption that polar clauses with the same polarity appear successively unless the context is changed with adversative expressions • collect candidate polar atoms with their tentative polarities as those adjacent to the polar clauses which have been identified by their domain-independent polar atoms in the initial lexicon • Intra-sentential and inter-sentential contexts to obtain more candidate polar atoms

  8. coherent precision and coherent density • coherent precision • coherent density • n=0 同一句, 1 隔一句 • Baseline所有positive比例

  9. Unsupervised Learning forAcquisition of Polar Atoms • We can assume • cd(d,L+a)約等於cd(d,L) • cp(d,L+a)約等於cp(d,L) when L is large • confidence interval using approximationwith the F-distribution

  10. Evaluation • Evaluation by Polar Atoms • two annotators evaluated 200 randomly selected candidate polar atoms in the digital camera domain • agreed upon in 89% of the cases and the Kappa value was 0.83 • Type Precision, Token Precision, Relative Recall • # of detected polar clauses with the expanded lexicon to the number of detected polar clauses with the initial lexicon

  11. Robustness for DifferentConditions • Diversity of Corpora • Size of the Initial Lexicon

More Related