Towards statistical paraphrase generation: preliminary evaluations of grammaticality

Stephen Wan; Mark Dras; Robert Dale; Cécile Paris

Towards statistical paraphrase generation: preliminary evaluations of grammaticality

Stephen Wan, Mark Dras, Robert Dale, Cécile Paris

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

Abstract

Summary sentences are often paraphrases of existing sentences. They may be made up of recycled fragments of text taken from important sentences in an input document. We investigate the use of a statistical sentence generation technique that recombines words probabilistically in order to create new sentences. Given a set of event-related sentences, we use an extended version of the Viterbi algorithm which employs dependency relation and bigram probabilities to find the most probable summary sentence. Using precision and recall metrics for verb arguments as a measure of grammaticality, we find that our system performs better than a bigram baseline, producing fewer spurious verb arguments.

Original language	English
Title of host publication	Proceedings of the Third International Workshop on Paraphrasing (IWP2005)
Editors	Mark Dras, Kazuhide Yamamoto
Place of Publication	Japan
Publisher	Asian Federation of Natural Language Processing
Pages	88-95
Number of pages	8
Publication status	Published - 2005
Event	International Workshop on Paraphrasing (3rd : 2005) - Jeju Island, South Korea Duration: 14 Oct 2005 → 14 Oct 2005

Workshop

Workshop	International Workshop on Paraphrasing (3rd : 2005)
City	Jeju Island, South Korea
Period	14/10/05 → 14/10/05

Cite this

@inproceedings{4f599e007ef842bd99b181ca0f7ac675,

title = "Towards statistical paraphrase generation: preliminary evaluations of grammaticality",

abstract = "Summary sentences are often paraphrases of existing sentences. They may be made up of recycled fragments of text taken from important sentences in an input document. We investigate the use of a statistical sentence generation technique that recombines words probabilistically in order to create new sentences. Given a set of event-related sentences, we use an extended version of the Viterbi algorithm which employs dependency relation and bigram probabilities to find the most probable summary sentence. Using precision and recall metrics for verb arguments as a measure of grammaticality, we find that our system performs better than a bigram baseline, producing fewer spurious verb arguments.",

author = "Stephen Wan and Mark Dras and Robert Dale and C{\'e}cile Paris",

year = "2005",

language = "English",

pages = "88--95",

editor = "Mark Dras and Kazuhide Yamamoto",

booktitle = "Proceedings of the Third International Workshop on Paraphrasing (IWP2005)",

publisher = "Asian Federation of Natural Language Processing",

note = "International Workshop on Paraphrasing (3rd : 2005) ; Conference date: 14-10-2005 Through 14-10-2005",

}

Wan, S, Dras, M , Dale, R & Paris, C 2005, Towards statistical paraphrase generation: preliminary evaluations of grammaticality. in M Dras & K Yamamoto (eds), Proceedings of the Third International Workshop on Paraphrasing (IWP2005). Asian Federation of Natural Language Processing, Japan, pp. 88-95, International Workshop on Paraphrasing (3rd : 2005), Jeju Island, South Korea, 14/10/05.

Towards statistical paraphrase generation: preliminary evaluations of grammaticality. / Wan, Stephen; Dras, Mark ; Dale, Robert et al.
Proceedings of the Third International Workshop on Paraphrasing (IWP2005). ed. / Mark Dras; Kazuhide Yamamoto. Japan: Asian Federation of Natural Language Processing, 2005. p. 88-95.

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

TY - GEN

T1 - Towards statistical paraphrase generation

T2 - International Workshop on Paraphrasing (3rd : 2005)

AU - Wan, Stephen

AU - Dras, Mark

AU - Dale, Robert

AU - Paris, Cécile

PY - 2005

Y1 - 2005

N2 - Summary sentences are often paraphrases of existing sentences. They may be made up of recycled fragments of text taken from important sentences in an input document. We investigate the use of a statistical sentence generation technique that recombines words probabilistically in order to create new sentences. Given a set of event-related sentences, we use an extended version of the Viterbi algorithm which employs dependency relation and bigram probabilities to find the most probable summary sentence. Using precision and recall metrics for verb arguments as a measure of grammaticality, we find that our system performs better than a bigram baseline, producing fewer spurious verb arguments.

AB - Summary sentences are often paraphrases of existing sentences. They may be made up of recycled fragments of text taken from important sentences in an input document. We investigate the use of a statistical sentence generation technique that recombines words probabilistically in order to create new sentences. Given a set of event-related sentences, we use an extended version of the Viterbi algorithm which employs dependency relation and bigram probabilities to find the most probable summary sentence. Using precision and recall metrics for verb arguments as a measure of grammaticality, we find that our system performs better than a bigram baseline, producing fewer spurious verb arguments.

UR - https://www.scopus.com/pages/publications/105017707230

M3 - Conference proceeding contribution

SP - 88

EP - 95

BT - Proceedings of the Third International Workshop on Paraphrasing (IWP2005)

A2 - Dras, Mark

A2 - Yamamoto, Kazuhide

PB - Asian Federation of Natural Language Processing

CY - Japan

Y2 - 14 October 2005 through 14 October 2005

ER -

Towards statistical paraphrase generation: preliminary evaluations of grammaticality

Abstract

Workshop

Other files and links

Fingerprint

Cite this