Improved extraction assessment through better language models

Arun Ahuja; Douglas C Downey

Improved extraction assessment through better language models

Arun Ahuja^*, Douglas C Downey

^*Corresponding author for this work

Computer Science

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

10 Scopus citations

Abstract

A variety of information extraction techniques rely on the fact that instances of the same relation are "distributionally similar," in that they tend to appear in similar textual contexts. We demonstrate that extraction accuracy depends heavily on the accuracy of the language model utilized to estimate distributional similarity. An unsupervised model selection technique based on this observation is shown to reduce extraction and type-checking error by 26% over previous results, in experiments with Hidden Markov Models. The results suggest that optimizing statistical language models over unlabeled data is a promising direction for improving weakly supervised and unsupervised information extraction.

Original language	English (US)
Title of host publication	NAACL HLT 2010 - Human Language Technologies
Subtitle of host publication	The 2010 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Proceedings of the Main Conference
Pages	225-228
Number of pages	4
State	Published - 2010
Event	2010 Human Language Technologies Conference ofthe North American Chapter of the Association for Computational Linguistics, NAACL HLT 2010 - Los Angeles, CA, United States Duration: Jun 2 2010 → Jun 4 2010

Publication series

Name	NAACL HLT 2010 - Human Language Technologies: The 2010 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Proceedings of the Main Conference

Other

Other	2010 Human Language Technologies Conference ofthe North American Chapter of the Association for Computational Linguistics, NAACL HLT 2010
Country/Territory	United States
City	Los Angeles, CA
Period	6/2/10 → 6/4/10

ASJC Scopus subject areas

Language and Linguistics
Linguistics and Language

Cite this

Ahuja, A., & Downey, D. C. (2010). Improved extraction assessment through better language models. In NAACL HLT 2010 - Human Language Technologies: The 2010 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Proceedings of the Main Conference (pp. 225-228). (NAACL HLT 2010 - Human Language Technologies: The 2010 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Proceedings of the Main Conference).

Ahuja, Arun ; Downey, Douglas C. / Improved extraction assessment through better language models. NAACL HLT 2010 - Human Language Technologies: The 2010 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Proceedings of the Main Conference. 2010. pp. 225-228 (NAACL HLT 2010 - Human Language Technologies: The 2010 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Proceedings of the Main Conference).

@inproceedings{8b8ebbb446d84c58bb8ceac33702285d,

title = "Improved extraction assessment through better language models",

abstract = "A variety of information extraction techniques rely on the fact that instances of the same relation are {"}distributionally similar,{"} in that they tend to appear in similar textual contexts. We demonstrate that extraction accuracy depends heavily on the accuracy of the language model utilized to estimate distributional similarity. An unsupervised model selection technique based on this observation is shown to reduce extraction and type-checking error by 26% over previous results, in experiments with Hidden Markov Models. The results suggest that optimizing statistical language models over unlabeled data is a promising direction for improving weakly supervised and unsupervised information extraction.",

author = "Arun Ahuja and Downey, {Douglas C}",

year = "2010",

language = "English (US)",

isbn = "1932432655",

series = "NAACL HLT 2010 - Human Language Technologies: The 2010 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Proceedings of the Main Conference",

pages = "225--228",

booktitle = "NAACL HLT 2010 - Human Language Technologies",

note = "2010 Human Language Technologies Conference ofthe North American Chapter of the Association for Computational Linguistics, NAACL HLT 2010 ; Conference date: 02-06-2010 Through 04-06-2010",

}

Ahuja, A & Downey, DC 2010, Improved extraction assessment through better language models. in NAACL HLT 2010 - Human Language Technologies: The 2010 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Proceedings of the Main Conference. NAACL HLT 2010 - Human Language Technologies: The 2010 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Proceedings of the Main Conference, pp. 225-228, 2010 Human Language Technologies Conference ofthe North American Chapter of the Association for Computational Linguistics, NAACL HLT 2010, Los Angeles, CA, United States, 6/2/10.

Improved extraction assessment through better language models. / Ahuja, Arun; Downey, Douglas C.
NAACL HLT 2010 - Human Language Technologies: The 2010 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Proceedings of the Main Conference. 2010. p. 225-228 (NAACL HLT 2010 - Human Language Technologies: The 2010 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Proceedings of the Main Conference).

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

TY - GEN

T1 - Improved extraction assessment through better language models

AU - Ahuja, Arun

AU - Downey, Douglas C

PY - 2010

Y1 - 2010

N2 - A variety of information extraction techniques rely on the fact that instances of the same relation are "distributionally similar," in that they tend to appear in similar textual contexts. We demonstrate that extraction accuracy depends heavily on the accuracy of the language model utilized to estimate distributional similarity. An unsupervised model selection technique based on this observation is shown to reduce extraction and type-checking error by 26% over previous results, in experiments with Hidden Markov Models. The results suggest that optimizing statistical language models over unlabeled data is a promising direction for improving weakly supervised and unsupervised information extraction.

AB - A variety of information extraction techniques rely on the fact that instances of the same relation are "distributionally similar," in that they tend to appear in similar textual contexts. We demonstrate that extraction accuracy depends heavily on the accuracy of the language model utilized to estimate distributional similarity. An unsupervised model selection technique based on this observation is shown to reduce extraction and type-checking error by 26% over previous results, in experiments with Hidden Markov Models. The results suggest that optimizing statistical language models over unlabeled data is a promising direction for improving weakly supervised and unsupervised information extraction.

UR - http://www.scopus.com/inward/record.url?scp=84858433233&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84858433233&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:84858433233

SN - 1932432655

SN - 9781932432657

T3 - NAACL HLT 2010 - Human Language Technologies: The 2010 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Proceedings of the Main Conference

SP - 225

EP - 228

BT - NAACL HLT 2010 - Human Language Technologies

T2 - 2010 Human Language Technologies Conference ofthe North American Chapter of the Association for Computational Linguistics, NAACL HLT 2010

Y2 - 2 June 2010 through 4 June 2010

ER -

Ahuja A, Downey DC. Improved extraction assessment through better language models. In NAACL HLT 2010 - Human Language Technologies: The 2010 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Proceedings of the Main Conference. 2010. p. 225-228. (NAACL HLT 2010 - Human Language Technologies: The 2010 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Proceedings of the Main Conference).

Improved extraction assessment through better language models

Abstract

Publication series

Other

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this