A method of Chinese coreference resolution combined multi-features in discourse

Shu Min Shi*, He Yan Huang, Rui Yang Chen

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

2 Citations (Scopus)

Abstract

Coreference that is a kind of ubiquitous language phenomenon makes the topic more highlighted and the narration more concise and coherent in discourse. Conversely, it leads to ambiguity in Natural Language Processing as well. Coreference resolution is the process that eliminates the indeterminacy caused by coreferential forms. To improve the current system, a method of coreference resolution combined with multi-features, mainly including clause's or full-sentence's distance, semantic class, and shorten-form features, is proposed in this paper. Experiments show that those features are valuable and have certain effects on the performance of resolution. It can be verified by both precision and F-measure in Chinese-oriented text discourse. In addition, a novelty point absorbed somewhat domain ontological idea in our previous work and embodied further in this paper is rather than usual manual linguistic rules-based approaches for processing the resolution.

Original languageEnglish
Title of host publication2010 International Conference on Machine Learning and Cybernetics, ICMLC 2010
Pages1311-1316
Number of pages6
DOIs
Publication statusPublished - 2010
Event2010 International Conference on Machine Learning and Cybernetics, ICMLC 2010 - Qingdao, China
Duration: 11 Jul 201014 Jul 2010

Publication series

Name2010 International Conference on Machine Learning and Cybernetics, ICMLC 2010
Volume3

Conference

Conference2010 International Conference on Machine Learning and Cybernetics, ICMLC 2010
Country/TerritoryChina
CityQingdao
Period11/07/1014/07/10

Keywords

  • CRF
  • Coreference resolution
  • Decision tree
  • Multi-features
  • NER

Fingerprint

Dive into the research topics of 'A method of Chinese coreference resolution combined multi-features in discourse'. Together they form a unique fingerprint.

Cite this