Thai text coherence structuring with coordinating and subordinating relations for text summarization

Authors:
Thana Sukvaree;Asanee Kawtrakul;Jean Caelen
Affiliations:
Department of Computer Engineering, Kasetsart University, Bangkok, Thailand;Department of Computer Engineering, Kasetsart University, Bangkok, Thailand;Laboratory CLIPS, University of Joseph Fourier, Grenoble Cedex 9, France
Venue:
CONTEXT'07 Proceedings of the 6th international and interdisciplinary conference on Modeling and using context
Year:
2007

Citing 8
Cited 0

A problem for RST: the need for multi-level discourse analysis

Computational Linguistics
Centering: a framework for modeling the local coherence of discourse

Computational Linguistics
New Methods in Automatic Extracting

Journal of the ACM (JACM)
A vector space model for automatic indexing

Communications of the ACM
The rhetorical parsing of natural language texts

ACL '98 Proceedings of the 35th Annual Meeting of the Association for Computational Linguistics and Eighth Conference of the European Chapter of the Association for Computational Linguistics
Automated text summarization and the SUMMARIST system

TIPSTER '98 Proceedings of a workshop on held at Baltimore, Maryland: October 13-15, 1998
The right frontier constraint as conditional

CICLing'05 Proceedings of the 6th international conference on Computational Linguistics and Intelligent Text Processing
Summarisation through discourse structure

CICLing'05 Proceedings of the 6th international conference on Computational Linguistics and Intelligent Text Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

Text summarization with the consideration of coherence can be achieved by using discourse processing with the Rhetorical Structure Theory (RST). Additional problems on relational ambiguity may arise, especially in Thai. For example, the use of cue words, i.e. "tae/???" (meaning "but"), can be identified as a contrast relation or an elaboration relation. Therefore, we propose the reduction of the ambiguity level by reducing the relation types to two, namely Coordinating and Subordinating relation. Our framework is to concentrate on coherence structuring which requires the following 3 steps: (1) identify an attachment point for an incoming discourse unit by using our Adaptive Right-frontier algorithm; (2) extract Coordinating and Subordinating relations through the identification of linguistic coherence features in the lexical and phrasal level, using Bayesian techniques; (3) construct coherence tree structures, The accuracy is 70.45% for the first step, 77.47% and 79.89% for COR and SUBR extraction respectively in the second step and 64.94% in constructing coherent tree of the third.