Combination strategies for semantic role labeling

Mihai Surdeanu, Lluís Marquez, Xavier Carreras, Pere R. Comas

Research output: Contribution to journalArticle

43 Citations (Scopus)

Abstract

This paper introduces and analyzes a battery of inference models for the problem of semantic role labeling: one based on constraint satisfaction, and several strategies that model the inference as a meta-learning problem using discriminative classifiers. These classifiers are developed with a rich set of novel features that encode proposition and sentence-level information. To our knowledge, this is the first work that: (a) performs a thorough analysis of learning-based inference models for semantic role labeling, and (b) compares several inference strategies in this context. We evaluate the proposed inference strategies in the framework of the CoNLL-2005 shared task using only automatically-generated syntactic information. The extensive experimental evaluation and analysis indicates that all the proposed inference strategies are successful -they all outperform the current best results reported in the CoNLL-2005 evaluation exercise- but each of the proposed approaches has its advantages and disadvantages. Several important traits of a state-of-the-art SRL combination strategy emerge from this analysis: (i) individual models should be combined at the granularity of candidate arguments rather than at the granularity of complete solutions; (ii) the best combination strategy uses an inference model based in learning; and (iii) the learning-based inference benefits from max-margin classifiers and global feedback.

Original languageEnglish (US)
Pages (from-to)105-151
Number of pages47
JournalJournal of Artificial Intelligence Research
Volume29
StatePublished - May 2007
Externally publishedYes

Fingerprint

Labeling
Semantics
Classifiers
Syntactics
Feedback

ASJC Scopus subject areas

  • Control and Systems Engineering
  • Artificial Intelligence

Cite this

Combination strategies for semantic role labeling. / Surdeanu, Mihai; Marquez, Lluís; Carreras, Xavier; Comas, Pere R.

In: Journal of Artificial Intelligence Research, Vol. 29, 05.2007, p. 105-151.

Research output: Contribution to journalArticle

Surdeanu, M, Marquez, L, Carreras, X & Comas, PR 2007, 'Combination strategies for semantic role labeling', Journal of Artificial Intelligence Research, vol. 29, pp. 105-151.
Surdeanu, Mihai ; Marquez, Lluís ; Carreras, Xavier ; Comas, Pere R. / Combination strategies for semantic role labeling. In: Journal of Artificial Intelligence Research. 2007 ; Vol. 29. pp. 105-151.
@article{dc14519370c84bebb8b8f71767f658fd,
title = "Combination strategies for semantic role labeling",
abstract = "This paper introduces and analyzes a battery of inference models for the problem of semantic role labeling: one based on constraint satisfaction, and several strategies that model the inference as a meta-learning problem using discriminative classifiers. These classifiers are developed with a rich set of novel features that encode proposition and sentence-level information. To our knowledge, this is the first work that: (a) performs a thorough analysis of learning-based inference models for semantic role labeling, and (b) compares several inference strategies in this context. We evaluate the proposed inference strategies in the framework of the CoNLL-2005 shared task using only automatically-generated syntactic information. The extensive experimental evaluation and analysis indicates that all the proposed inference strategies are successful -they all outperform the current best results reported in the CoNLL-2005 evaluation exercise- but each of the proposed approaches has its advantages and disadvantages. Several important traits of a state-of-the-art SRL combination strategy emerge from this analysis: (i) individual models should be combined at the granularity of candidate arguments rather than at the granularity of complete solutions; (ii) the best combination strategy uses an inference model based in learning; and (iii) the learning-based inference benefits from max-margin classifiers and global feedback.",
author = "Mihai Surdeanu and Llu{\'i}s Marquez and Xavier Carreras and Comas, {Pere R.}",
year = "2007",
month = "5",
language = "English (US)",
volume = "29",
pages = "105--151",
journal = "Journal of Artificial Intelligence Research",
issn = "1076-9757",
publisher = "Morgan Kaufmann Publishers, Inc.",

}

TY - JOUR

T1 - Combination strategies for semantic role labeling

AU - Surdeanu, Mihai

AU - Marquez, Lluís

AU - Carreras, Xavier

AU - Comas, Pere R.

PY - 2007/5

Y1 - 2007/5

N2 - This paper introduces and analyzes a battery of inference models for the problem of semantic role labeling: one based on constraint satisfaction, and several strategies that model the inference as a meta-learning problem using discriminative classifiers. These classifiers are developed with a rich set of novel features that encode proposition and sentence-level information. To our knowledge, this is the first work that: (a) performs a thorough analysis of learning-based inference models for semantic role labeling, and (b) compares several inference strategies in this context. We evaluate the proposed inference strategies in the framework of the CoNLL-2005 shared task using only automatically-generated syntactic information. The extensive experimental evaluation and analysis indicates that all the proposed inference strategies are successful -they all outperform the current best results reported in the CoNLL-2005 evaluation exercise- but each of the proposed approaches has its advantages and disadvantages. Several important traits of a state-of-the-art SRL combination strategy emerge from this analysis: (i) individual models should be combined at the granularity of candidate arguments rather than at the granularity of complete solutions; (ii) the best combination strategy uses an inference model based in learning; and (iii) the learning-based inference benefits from max-margin classifiers and global feedback.

AB - This paper introduces and analyzes a battery of inference models for the problem of semantic role labeling: one based on constraint satisfaction, and several strategies that model the inference as a meta-learning problem using discriminative classifiers. These classifiers are developed with a rich set of novel features that encode proposition and sentence-level information. To our knowledge, this is the first work that: (a) performs a thorough analysis of learning-based inference models for semantic role labeling, and (b) compares several inference strategies in this context. We evaluate the proposed inference strategies in the framework of the CoNLL-2005 shared task using only automatically-generated syntactic information. The extensive experimental evaluation and analysis indicates that all the proposed inference strategies are successful -they all outperform the current best results reported in the CoNLL-2005 evaluation exercise- but each of the proposed approaches has its advantages and disadvantages. Several important traits of a state-of-the-art SRL combination strategy emerge from this analysis: (i) individual models should be combined at the granularity of candidate arguments rather than at the granularity of complete solutions; (ii) the best combination strategy uses an inference model based in learning; and (iii) the learning-based inference benefits from max-margin classifiers and global feedback.

UR - http://www.scopus.com/inward/record.url?scp=34748813019&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=34748813019&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:34748813019

VL - 29

SP - 105

EP - 151

JO - Journal of Artificial Intelligence Research

JF - Journal of Artificial Intelligence Research

SN - 1076-9757

ER -