Robustness and generalization of role sets

Propbank vs. VerbNet

Bẽnat Zapirain, Eneko Agirre, Lluis Marques

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

9 Citations (Scopus)

Abstract

This paper presents an empirical study on the robustness and generalization of two alternative role sets for semantic role labeling: Prop- Bank numbered roles and VerbNet thematic roles. By testing a state-of-the-art SRL system with the two alternative role annotations, we show that the PropBank role set is more robust to the lack of verb-specific semantic information and generalizes better to infrequent and unseen predicates. Keeping in mind that thematic roles are better for application needs, we also tested the best way to generate VerbNet annotation. We conclude that tagging first PropBank roles and mapping into Verb- Net roles is as effective as training and tagging directly on VerbNet, and more robust for domain shifts.

Original languageEnglish
Title of host publicationACL-08: HLT - 46th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, Proceedings of the Conference
Pages550-558
Number of pages9
Publication statusPublished - 1 Dec 2008
Externally publishedYes
Event46th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, ACL-08: HLT - Columbus, OH, United States
Duration: 15 Jun 200820 Jun 2008

Other

Other46th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, ACL-08: HLT
CountryUnited States
CityColumbus, OH
Period15/6/0820/6/08

Fingerprint

Semantics
semantics
Labeling
bank
lack
Testing
Tagging
Verbs
Annotation
Robustness
Thematic Roles
Semantic Roles
Semantic Information
Props
Empirical Study

ASJC Scopus subject areas

  • Language and Linguistics
  • Computer Networks and Communications
  • Linguistics and Language

Cite this

Zapirain, B., Agirre, E., & Marques, L. (2008). Robustness and generalization of role sets: Propbank vs. VerbNet. In ACL-08: HLT - 46th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, Proceedings of the Conference (pp. 550-558)

Robustness and generalization of role sets : Propbank vs. VerbNet. / Zapirain, Bẽnat; Agirre, Eneko; Marques, Lluis.

ACL-08: HLT - 46th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, Proceedings of the Conference. 2008. p. 550-558.

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

Zapirain, B, Agirre, E & Marques, L 2008, Robustness and generalization of role sets: Propbank vs. VerbNet. in ACL-08: HLT - 46th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, Proceedings of the Conference. pp. 550-558, 46th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, ACL-08: HLT, Columbus, OH, United States, 15/6/08.
Zapirain B, Agirre E, Marques L. Robustness and generalization of role sets: Propbank vs. VerbNet. In ACL-08: HLT - 46th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, Proceedings of the Conference. 2008. p. 550-558
Zapirain, Bẽnat ; Agirre, Eneko ; Marques, Lluis. / Robustness and generalization of role sets : Propbank vs. VerbNet. ACL-08: HLT - 46th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, Proceedings of the Conference. 2008. pp. 550-558
@inproceedings{9715a69fd53c4a53818eab19fd6c9d5a,
title = "Robustness and generalization of role sets: Propbank vs. VerbNet",
abstract = "This paper presents an empirical study on the robustness and generalization of two alternative role sets for semantic role labeling: Prop- Bank numbered roles and VerbNet thematic roles. By testing a state-of-the-art SRL system with the two alternative role annotations, we show that the PropBank role set is more robust to the lack of verb-specific semantic information and generalizes better to infrequent and unseen predicates. Keeping in mind that thematic roles are better for application needs, we also tested the best way to generate VerbNet annotation. We conclude that tagging first PropBank roles and mapping into Verb- Net roles is as effective as training and tagging directly on VerbNet, and more robust for domain shifts.",
author = "Bẽnat Zapirain and Eneko Agirre and Lluis Marques",
year = "2008",
month = "12",
day = "1",
language = "English",
isbn = "9781932432046",
pages = "550--558",
booktitle = "ACL-08: HLT - 46th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, Proceedings of the Conference",

}

TY - GEN

T1 - Robustness and generalization of role sets

T2 - Propbank vs. VerbNet

AU - Zapirain, Bẽnat

AU - Agirre, Eneko

AU - Marques, Lluis

PY - 2008/12/1

Y1 - 2008/12/1

N2 - This paper presents an empirical study on the robustness and generalization of two alternative role sets for semantic role labeling: Prop- Bank numbered roles and VerbNet thematic roles. By testing a state-of-the-art SRL system with the two alternative role annotations, we show that the PropBank role set is more robust to the lack of verb-specific semantic information and generalizes better to infrequent and unseen predicates. Keeping in mind that thematic roles are better for application needs, we also tested the best way to generate VerbNet annotation. We conclude that tagging first PropBank roles and mapping into Verb- Net roles is as effective as training and tagging directly on VerbNet, and more robust for domain shifts.

AB - This paper presents an empirical study on the robustness and generalization of two alternative role sets for semantic role labeling: Prop- Bank numbered roles and VerbNet thematic roles. By testing a state-of-the-art SRL system with the two alternative role annotations, we show that the PropBank role set is more robust to the lack of verb-specific semantic information and generalizes better to infrequent and unseen predicates. Keeping in mind that thematic roles are better for application needs, we also tested the best way to generate VerbNet annotation. We conclude that tagging first PropBank roles and mapping into Verb- Net roles is as effective as training and tagging directly on VerbNet, and more robust for domain shifts.

UR - http://www.scopus.com/inward/record.url?scp=78650461326&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=78650461326&partnerID=8YFLogxK

M3 - Conference contribution

SN - 9781932432046

SP - 550

EP - 558

BT - ACL-08: HLT - 46th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, Proceedings of the Conference

ER -