Learning sentiment classification model from labeled features

Yulan He

Research output: Chapter in Book/Report/Conference proceedingOther chapter contribution

Abstract

We propose a novel framework where an initial classifier is learned by incorporating prior information extracted from an existing sentiment lexicon. Preferences on expectations of sentiment labels of those lexicon words are expressed using generalized expectation criteria. Documents classified with high confidence are then used as pseudo-labeled examples for automatical domain-specific feature acquisition. The word-class distributions of such self-learned features are estimated from the pseudo-labeled examples and are used to train another classifier by constraining the model's predictions on unlabeled instances. Experiments on both the movie review data and the multi-domain sentiment dataset show that our approach attains comparable or better performance than exiting weakly-supervised sentiment classification methods despite using no labeled documents.
Original languageEnglish
Title of host publicationProceeding : CIKM '10 proceedings of the 19th ACM international conference on information and knowledge management
Place of PublicationNew York (US)
PublisherACM
Pages1685-1688
Number of pages4
ISBN (Print)978-1-4503-0099-5
DOIs
Publication statusPublished - 2010
Event19th ACM international conference on information and knowledge management, CIKM '10 - Toronto, Canada
Duration: 26 Oct 201030 Oct 2010

Conference

Conference19th ACM international conference on information and knowledge management, CIKM '10
CountryCanada
CityToronto
Period26/10/1030/10/10

Fingerprint

Classifiers
Labels
Experiments

Keywords

  • sentiment analysis
  • opinion mining
  • generalized expectation
  • self-learned features
  • weakly-supervised classification

Cite this

He, Y. (2010). Learning sentiment classification model from labeled features. In Proceeding : CIKM '10 proceedings of the 19th ACM international conference on information and knowledge management (pp. 1685-1688). New York (US): ACM. https://doi.org/10.1145/1871437.1871704
He, Yulan. / Learning sentiment classification model from labeled features. Proceeding : CIKM '10 proceedings of the 19th ACM international conference on information and knowledge management. New York (US) : ACM, 2010. pp. 1685-1688
@inbook{d4beb72b3856462a8b0e53882422b5b4,
title = "Learning sentiment classification model from labeled features",
abstract = "We propose a novel framework where an initial classifier is learned by incorporating prior information extracted from an existing sentiment lexicon. Preferences on expectations of sentiment labels of those lexicon words are expressed using generalized expectation criteria. Documents classified with high confidence are then used as pseudo-labeled examples for automatical domain-specific feature acquisition. The word-class distributions of such self-learned features are estimated from the pseudo-labeled examples and are used to train another classifier by constraining the model's predictions on unlabeled instances. Experiments on both the movie review data and the multi-domain sentiment dataset show that our approach attains comparable or better performance than exiting weakly-supervised sentiment classification methods despite using no labeled documents.",
keywords = "sentiment analysis, opinion mining, generalized expectation, self-learned features, weakly-supervised classification",
author = "Yulan He",
year = "2010",
doi = "10.1145/1871437.1871704",
language = "English",
isbn = "978-1-4503-0099-5",
pages = "1685--1688",
booktitle = "Proceeding : CIKM '10 proceedings of the 19th ACM international conference on information and knowledge management",
publisher = "ACM",
address = "United States",

}

He, Y 2010, Learning sentiment classification model from labeled features. in Proceeding : CIKM '10 proceedings of the 19th ACM international conference on information and knowledge management. ACM, New York (US), pp. 1685-1688, 19th ACM international conference on information and knowledge management, CIKM '10, Toronto, Canada, 26/10/10. https://doi.org/10.1145/1871437.1871704

Learning sentiment classification model from labeled features. / He, Yulan.

Proceeding : CIKM '10 proceedings of the 19th ACM international conference on information and knowledge management. New York (US) : ACM, 2010. p. 1685-1688.

Research output: Chapter in Book/Report/Conference proceedingOther chapter contribution

TY - CHAP

T1 - Learning sentiment classification model from labeled features

AU - He, Yulan

PY - 2010

Y1 - 2010

N2 - We propose a novel framework where an initial classifier is learned by incorporating prior information extracted from an existing sentiment lexicon. Preferences on expectations of sentiment labels of those lexicon words are expressed using generalized expectation criteria. Documents classified with high confidence are then used as pseudo-labeled examples for automatical domain-specific feature acquisition. The word-class distributions of such self-learned features are estimated from the pseudo-labeled examples and are used to train another classifier by constraining the model's predictions on unlabeled instances. Experiments on both the movie review data and the multi-domain sentiment dataset show that our approach attains comparable or better performance than exiting weakly-supervised sentiment classification methods despite using no labeled documents.

AB - We propose a novel framework where an initial classifier is learned by incorporating prior information extracted from an existing sentiment lexicon. Preferences on expectations of sentiment labels of those lexicon words are expressed using generalized expectation criteria. Documents classified with high confidence are then used as pseudo-labeled examples for automatical domain-specific feature acquisition. The word-class distributions of such self-learned features are estimated from the pseudo-labeled examples and are used to train another classifier by constraining the model's predictions on unlabeled instances. Experiments on both the movie review data and the multi-domain sentiment dataset show that our approach attains comparable or better performance than exiting weakly-supervised sentiment classification methods despite using no labeled documents.

KW - sentiment analysis

KW - opinion mining

KW - generalized expectation

KW - self-learned features

KW - weakly-supervised classification

UR - http://www.scopus.com/inward/record.url?scp=78651271277&partnerID=8YFLogxK

UR - http://dl.acm.org/citation.cfm?id=1871704

U2 - 10.1145/1871437.1871704

DO - 10.1145/1871437.1871704

M3 - Other chapter contribution

AN - SCOPUS:78651271277

SN - 978-1-4503-0099-5

SP - 1685

EP - 1688

BT - Proceeding : CIKM '10 proceedings of the 19th ACM international conference on information and knowledge management

PB - ACM

CY - New York (US)

ER -

He Y. Learning sentiment classification model from labeled features. In Proceeding : CIKM '10 proceedings of the 19th ACM international conference on information and knowledge management. New York (US): ACM. 2010. p. 1685-1688 https://doi.org/10.1145/1871437.1871704