Large Margin Synchronous Generation and its Application to Sentence Compression

Trevor Cohn; Mirella Lapata

Large Margin Synchronous Generation and its Application to Sentence Compression

School of Informatics

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

This paper presents a tree-to-tree transduction method for text rewriting. Our model is based on synchronous tree substitution grammar, a formalism that allows local distortion of the tree topology and can thus naturally capture structural mismatches. We describe an algorithm for decoding in this framework and show how the model can be trained discriminatively within a large margin framework. Experimental results on sentence compression bring significant improvements over a state-of-the-art model.

Original language	English
Title of host publication	Proceedings of the 2007 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning
Publisher	Association for Computational Linguistics
Pages	73-82
Number of pages	10
Publication status	Published - 2007

Access to Document

2007_Cohn_Lapata_Large Margin Synchronous Genreation and its Application to Sentence CompressionFinal published version, 187 KB

http://www.aclweb.org/anthology/D07-1008

Application-based Text-to-Text Generation
Lapata, M. (Principal Investigator)
EPSRC
31/03/06 → 30/03/09
Project: Research
Statistical model for text-to-text generation
Lapata, M. (Principal Investigator)
EPSRC
1/02/05 → 28/02/11
Project: Research

Cite this

@inproceedings{04c53378040b4f08b2bcf8d1b5fff0df,

title = "Large Margin Synchronous Generation and its Application to Sentence Compression",

abstract = "This paper presents a tree-to-tree transduction method for text rewriting. Our model is based on synchronous tree substitution grammar, a formalism that allows local distortion of the tree topology and can thus naturally capture structural mismatches. We describe an algorithm for decoding in this framework and show how the model can be trained discriminatively within a large margin framework. Experimental results on sentence compression bring significant improvements over a state-of-the-art model.",

author = "Trevor Cohn and Mirella Lapata",

year = "2007",

language = "English",

pages = "73--82",

booktitle = "Proceedings of the 2007 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning",

publisher = "Association for Computational Linguistics",

}

Large Margin Synchronous Generation and its Application to Sentence Compression. / Cohn, Trevor; Lapata, Mirella.
Proceedings of the 2007 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning. Association for Computational Linguistics, 2007. p. 73-82.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - Large Margin Synchronous Generation and its Application to Sentence Compression

AU - Cohn, Trevor

AU - Lapata, Mirella

PY - 2007

Y1 - 2007

N2 - This paper presents a tree-to-tree transduction method for text rewriting. Our model is based on synchronous tree substitution grammar, a formalism that allows local distortion of the tree topology and can thus naturally capture structural mismatches. We describe an algorithm for decoding in this framework and show how the model can be trained discriminatively within a large margin framework. Experimental results on sentence compression bring significant improvements over a state-of-the-art model.

AB - This paper presents a tree-to-tree transduction method for text rewriting. Our model is based on synchronous tree substitution grammar, a formalism that allows local distortion of the tree topology and can thus naturally capture structural mismatches. We describe an algorithm for decoding in this framework and show how the model can be trained discriminatively within a large margin framework. Experimental results on sentence compression bring significant improvements over a state-of-the-art model.

M3 - Conference contribution

SP - 73

EP - 82

BT - Proceedings of the 2007 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning

PB - Association for Computational Linguistics

ER -

Large Margin Synchronous Generation and its Application to Sentence Compression

Abstract

Access to Document

Fingerprint

Projects

Application-based Text-to-Text Generation

Statistical model for text-to-text generation

Cite this