Empirical evaluation of CRF-based bibliography extraction from reference strings

Manabu Ohta, Daiki Arauchi, Atsuhiro Takasu, Jun Adachi

Research output: Chapter in Book/Report/Conference proceedingConference contribution

4 Citations (Scopus)

Abstract

This paper reports an empirical evaluation of a CRF-based bibliography parser we have developed for reference strings of research papers. The parser uses a conditional random field (CRF) to estimate the correct bibliographic label such as an author's name and a title for each token in a reference string. We applied the parser specifically designed for reference strings to three academic journals, an English one and two Japanese ones, published in Japan. Experiments showed i) the parser correctly parsed from 90% to 94% of reference strings depending on the kinds of journals used and ii) segmentation errors induced by tokenization considerably degraded the final parsing accuracies. This paper also discusses some future directions of the bibliography extraction based on a detailed analysis of the experiments.

Original languageEnglish
Title of host publicationProceedings - 11th IAPR International Workshop on Document Analysis Systems, DAS 2014
PublisherIEEE Computer Society
Pages287-292
Number of pages6
ISBN (Print)9781479932436
DOIs
Publication statusPublished - 2014
Event11th IAPR International Workshop on Document Analysis Systems, DAS 2014 - Tours, France
Duration: Apr 7 2014Apr 10 2014

Other

Other11th IAPR International Workshop on Document Analysis Systems, DAS 2014
CountryFrance
CityTours
Period4/7/144/10/14

Fingerprint

Bibliographies
Labels
Experiments

Keywords

  • Bibliography extraction
  • Citation parsing
  • Conditional random field
  • Evaluation
  • Metadata

ASJC Scopus subject areas

  • Software

Cite this

Ohta, M., Arauchi, D., Takasu, A., & Adachi, J. (2014). Empirical evaluation of CRF-based bibliography extraction from reference strings. In Proceedings - 11th IAPR International Workshop on Document Analysis Systems, DAS 2014 (pp. 287-292). [6831015] IEEE Computer Society. https://doi.org/10.1109/DAS.2014.64

Empirical evaluation of CRF-based bibliography extraction from reference strings. / Ohta, Manabu; Arauchi, Daiki; Takasu, Atsuhiro; Adachi, Jun.

Proceedings - 11th IAPR International Workshop on Document Analysis Systems, DAS 2014. IEEE Computer Society, 2014. p. 287-292 6831015.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Ohta, M, Arauchi, D, Takasu, A & Adachi, J 2014, Empirical evaluation of CRF-based bibliography extraction from reference strings. in Proceedings - 11th IAPR International Workshop on Document Analysis Systems, DAS 2014., 6831015, IEEE Computer Society, pp. 287-292, 11th IAPR International Workshop on Document Analysis Systems, DAS 2014, Tours, France, 4/7/14. https://doi.org/10.1109/DAS.2014.64
Ohta M, Arauchi D, Takasu A, Adachi J. Empirical evaluation of CRF-based bibliography extraction from reference strings. In Proceedings - 11th IAPR International Workshop on Document Analysis Systems, DAS 2014. IEEE Computer Society. 2014. p. 287-292. 6831015 https://doi.org/10.1109/DAS.2014.64
Ohta, Manabu ; Arauchi, Daiki ; Takasu, Atsuhiro ; Adachi, Jun. / Empirical evaluation of CRF-based bibliography extraction from reference strings. Proceedings - 11th IAPR International Workshop on Document Analysis Systems, DAS 2014. IEEE Computer Society, 2014. pp. 287-292
@inproceedings{53d03f55dc18426fa8c7590bfe9b807a,
title = "Empirical evaluation of CRF-based bibliography extraction from reference strings",
abstract = "This paper reports an empirical evaluation of a CRF-based bibliography parser we have developed for reference strings of research papers. The parser uses a conditional random field (CRF) to estimate the correct bibliographic label such as an author's name and a title for each token in a reference string. We applied the parser specifically designed for reference strings to three academic journals, an English one and two Japanese ones, published in Japan. Experiments showed i) the parser correctly parsed from 90{\%} to 94{\%} of reference strings depending on the kinds of journals used and ii) segmentation errors induced by tokenization considerably degraded the final parsing accuracies. This paper also discusses some future directions of the bibliography extraction based on a detailed analysis of the experiments.",
keywords = "Bibliography extraction, Citation parsing, Conditional random field, Evaluation, Metadata",
author = "Manabu Ohta and Daiki Arauchi and Atsuhiro Takasu and Jun Adachi",
year = "2014",
doi = "10.1109/DAS.2014.64",
language = "English",
isbn = "9781479932436",
pages = "287--292",
booktitle = "Proceedings - 11th IAPR International Workshop on Document Analysis Systems, DAS 2014",
publisher = "IEEE Computer Society",
address = "United States",

}

TY - GEN

T1 - Empirical evaluation of CRF-based bibliography extraction from reference strings

AU - Ohta, Manabu

AU - Arauchi, Daiki

AU - Takasu, Atsuhiro

AU - Adachi, Jun

PY - 2014

Y1 - 2014

N2 - This paper reports an empirical evaluation of a CRF-based bibliography parser we have developed for reference strings of research papers. The parser uses a conditional random field (CRF) to estimate the correct bibliographic label such as an author's name and a title for each token in a reference string. We applied the parser specifically designed for reference strings to three academic journals, an English one and two Japanese ones, published in Japan. Experiments showed i) the parser correctly parsed from 90% to 94% of reference strings depending on the kinds of journals used and ii) segmentation errors induced by tokenization considerably degraded the final parsing accuracies. This paper also discusses some future directions of the bibliography extraction based on a detailed analysis of the experiments.

AB - This paper reports an empirical evaluation of a CRF-based bibliography parser we have developed for reference strings of research papers. The parser uses a conditional random field (CRF) to estimate the correct bibliographic label such as an author's name and a title for each token in a reference string. We applied the parser specifically designed for reference strings to three academic journals, an English one and two Japanese ones, published in Japan. Experiments showed i) the parser correctly parsed from 90% to 94% of reference strings depending on the kinds of journals used and ii) segmentation errors induced by tokenization considerably degraded the final parsing accuracies. This paper also discusses some future directions of the bibliography extraction based on a detailed analysis of the experiments.

KW - Bibliography extraction

KW - Citation parsing

KW - Conditional random field

KW - Evaluation

KW - Metadata

UR - http://www.scopus.com/inward/record.url?scp=84903744533&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84903744533&partnerID=8YFLogxK

U2 - 10.1109/DAS.2014.64

DO - 10.1109/DAS.2014.64

M3 - Conference contribution

SN - 9781479932436

SP - 287

EP - 292

BT - Proceedings - 11th IAPR International Workshop on Document Analysis Systems, DAS 2014

PB - IEEE Computer Society

ER -