Mining Chinese polarity shifters

Research output: Chapter in book / Conference proceedingConference article published in proceeding or bookAcademic researchpeer-review

Abstract

In sentiment analysis, polarity shifting means to shift the polarity of a sentiment clue. Compared with other natural language processing (NLP) tasks, to extract polarity shifters (polarity shifting patterns) in corpora is a challenging one, since the polarity shifters sometimes are very subtle, which often invalidates fully automatic approaches. In this paper, aiming to extract polarity shifters that invert or attenuate polarity, we use a semi-automatic approach based on pattern mining. The approach can greatly reduce the human annotating cost and cover as many polarity shifters as possible. We tested this approach on domain corpora, and encouraging experimental results are reported.
Original languageEnglish
Title of host publicationChinese Lexical Semantics - 16th Workshop, CLSW 2015, Revised Selected Papers
PublisherSpringer Verlag
Pages244-251
Number of pages8
ISBN (Print)9783319271934
DOIs
Publication statusPublished - 1 Jan 2015
Event16th Workshop on Chinese Lexical Semantics Workshop, CLSW 2015 - Beijing, China
Duration: 9 May 201511 May 2015

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9332
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference16th Workshop on Chinese Lexical Semantics Workshop, CLSW 2015
CountryChina
CityBeijing
Period9/05/1511/05/15

Keywords

  • Polarity shifting
  • Prior polarity
  • Sentiment analysis
  • Sequence mining

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this