Latent trees for coreference resolution

Research output: Journal contributionsJournal articlesResearchpeer-review

Standard

Latent trees for coreference resolution. / Fernandes, Eraldo Rezende; dos Santos, Cícero Nogueira; Milidiú, Ruy Luiz.
In: Computational Linguistics, Vol. 40, No. 4, 19.12.2014, p. 801-835.

Research output: Journal contributionsJournal articlesResearchpeer-review

Harvard

APA

Vancouver

Fernandes ER, dos Santos CN, Milidiú RL. Latent trees for coreference resolution. Computational Linguistics. 2014 Dec 19;40(4):801-835. doi: 10.1162/COLI_a_00200

Bibtex

@article{77fc8ee145124047b4766a9cd67a110b,
title = "Latent trees for coreference resolution",
abstract = "We describe a structure learning system for unrestricted coreference resolution that explores two key modeling techniques: latent coreference trees and automatic entropy-guided feature induction. The latent tree modeling makes the learning problem computationally feasible because it incorporates a meaningful hidden structure. Additionally, using an automatic feature induction method, we can efficiently build enhanced nonlinear models using linear model learning algorithms. We present empirical results that highlight the contribution of each modeling technique used in the proposed system. Empirical evaluation is performed on the multilingual unrestricted coreference CoNLL-2012 Shared Task data sets, which comprise three languages: Arabic, Chinese, and English. We apply the same system to all languages, except for minor adaptations to some language-dependent features such as nested mentions and specific static pronoun lists. A previous version of this system was submitted to the CoNLL-2012 Shared Task closed track, achieving an official score of 58:69, the best among the competitors. The unique enhancement added to the current system version is the inclusion of candidate arcs linking nested mentions for the Chinese language. By including such arcs, the score increases by almost 4.5 points for that language. The current system shows a score of 60:15, which corresponds to a 3:5% error reduction, and is the best performing system for each of the three languages.",
keywords = "Informatics, Business informatics",
author = "Fernandes, {Eraldo Rezende} and {dos Santos}, {C{\'i}cero Nogueira} and Milidi{\'u}, {Ruy Luiz}",
year = "2014",
month = dec,
day = "19",
doi = "10.1162/COLI_a_00200",
language = "English",
volume = "40",
pages = "801--835",
journal = "Computational Linguistics",
issn = "0891-2017",
publisher = "MIT Press",
number = "4",

}

RIS

TY - JOUR

T1 - Latent trees for coreference resolution

AU - Fernandes, Eraldo Rezende

AU - dos Santos, Cícero Nogueira

AU - Milidiú, Ruy Luiz

PY - 2014/12/19

Y1 - 2014/12/19

N2 - We describe a structure learning system for unrestricted coreference resolution that explores two key modeling techniques: latent coreference trees and automatic entropy-guided feature induction. The latent tree modeling makes the learning problem computationally feasible because it incorporates a meaningful hidden structure. Additionally, using an automatic feature induction method, we can efficiently build enhanced nonlinear models using linear model learning algorithms. We present empirical results that highlight the contribution of each modeling technique used in the proposed system. Empirical evaluation is performed on the multilingual unrestricted coreference CoNLL-2012 Shared Task data sets, which comprise three languages: Arabic, Chinese, and English. We apply the same system to all languages, except for minor adaptations to some language-dependent features such as nested mentions and specific static pronoun lists. A previous version of this system was submitted to the CoNLL-2012 Shared Task closed track, achieving an official score of 58:69, the best among the competitors. The unique enhancement added to the current system version is the inclusion of candidate arcs linking nested mentions for the Chinese language. By including such arcs, the score increases by almost 4.5 points for that language. The current system shows a score of 60:15, which corresponds to a 3:5% error reduction, and is the best performing system for each of the three languages.

AB - We describe a structure learning system for unrestricted coreference resolution that explores two key modeling techniques: latent coreference trees and automatic entropy-guided feature induction. The latent tree modeling makes the learning problem computationally feasible because it incorporates a meaningful hidden structure. Additionally, using an automatic feature induction method, we can efficiently build enhanced nonlinear models using linear model learning algorithms. We present empirical results that highlight the contribution of each modeling technique used in the proposed system. Empirical evaluation is performed on the multilingual unrestricted coreference CoNLL-2012 Shared Task data sets, which comprise three languages: Arabic, Chinese, and English. We apply the same system to all languages, except for minor adaptations to some language-dependent features such as nested mentions and specific static pronoun lists. A previous version of this system was submitted to the CoNLL-2012 Shared Task closed track, achieving an official score of 58:69, the best among the competitors. The unique enhancement added to the current system version is the inclusion of candidate arcs linking nested mentions for the Chinese language. By including such arcs, the score increases by almost 4.5 points for that language. The current system shows a score of 60:15, which corresponds to a 3:5% error reduction, and is the best performing system for each of the three languages.

KW - Informatics

KW - Business informatics

UR - http://www.scopus.com/inward/record.url?scp=84918531827&partnerID=8YFLogxK

U2 - 10.1162/COLI_a_00200

DO - 10.1162/COLI_a_00200

M3 - Journal articles

AN - SCOPUS:84918531827

VL - 40

SP - 801

EP - 835

JO - Computational Linguistics

JF - Computational Linguistics

SN - 0891-2017

IS - 4

ER -

DOI

Recently viewed

Publications

  1. Toxicity testing with luminescent bacteria - Characterization of an automated method for the combined assessment of acute and chronic effects
  2. RelHunter
  3. Swarm Robotics, or: The Smartness of 'a bunch of cheap dumb things'
  4. Perceptions of Organizational Downsizing
  5. Policy implementation through multi-level governance
  6. Pre-service mathematics teachers' modelling processes within model eliciting activity through digital technologies
  7. Advantages and difficulties of conducting thinking-aloud protocols in the school setting
  8. Development of a procedure for forming assisted thermal joining of tubes
  9. The complementarity of single-species and ecosystem-oriented research in conservation research
  10. Innovation in Continuing Engineering Education with focus on gender and non-traditional students' pathways
  11. Does transition to IFRS substantially affect key financial ratios in shareholder-oriented common law regimes?
  12. Do it again
  13. Classification of playing position in elite junior Australian football using technical skill indicators
  14. Global patterns of ecologically unequal exchange
  15. The use of force against terrorists
  16. Wir sind ihr
  17. Delivering community benefits through REDD plus : Lessons from Joint Forest Management in Zambia
  18. Internet-Based Prevention of Depression in Employees
  19. Toward a Production-Oriented Imagology
  20. The Computational Turn, or, a New Weltbild
  21. Archival research on carbon reporting quality. A review of determinants and consequences for firm value
  22. Community and Training in NFDI4DS
  23. Kriminalisierung und Versicherheitlichung von Migration. Editorial
  24. Assoggettamento/Soggettivazione
  25. On the micro-structure of the German export boom
  26. The Measurement of Grip-Strength in Automobiles
  27. Front in the mouth, front in the word
  28. Intra- and interspecific hybridization in invasive Siberian elm