Tree kernels for semantic role labeling

Alessandro Moschitti, Daniele Pighin, Roberto Basili

Research output: Contribution to journalArticle

103 Citations (Scopus)

Abstract

The availability of large scale data sets of manually annotated predicate-argument structures has recently favored the use of machine learning approaches to the design of automated semantic role labeling (SRL) systems. The main research in this area relates to the design choices for feature representation and for effective decompositions of the task in different learning models. Regarding the former choice, structural properties of full syntactic parses are largely employed as they represent ways to encode different principles suggested by the linking theory between syntax and semantics. The latter choice relates to several learning schemes over global views of the parses. For example, re-ranking stages operating over alternative predicate-argument sequences of the same sentence have shown to be very effective. In this article, we propose several kernel functions to model parse tree properties in kernelbased machines, for example, perceptrons or support vector machines. In particular, we define different kinds of tree kernels as general approaches to feature engineering in SRL. Moreover, we extensively experiment with such kernels to investigate their contribution to individual stages of an SRL architecture both in isolation and in combination with other traditional manually coded features. The results for boundary recognition, classification, and re-ranking stages provide systematic evidence about the significant impact of tree kernels on the overall accuracy, especially when the amount of training data is small. As a conclusive result, tree kernels allow for a general and easily portable feature engineering method which is applicable to a large family of natural language processing tasks.

Original languageEnglish
Pages (from-to)193-224
Number of pages32
JournalComputational Linguistics
Volume34
Issue number2
DOIs
Publication statusPublished - 1 Jun 2008
Externally publishedYes

Fingerprint

Labeling
Semantics
semantics
syntax theory
ranking
engineering
learning
large family
Syntactics
Support vector machines
Learning systems
Structural properties
social isolation
Availability
Neural networks
Decomposition
Semantic Roles
Kernel
experiment
Processing

ASJC Scopus subject areas

  • Computer Science Applications
  • Computational Theory and Mathematics
  • Linguistics and Language
  • Language and Linguistics

Cite this

Tree kernels for semantic role labeling. / Moschitti, Alessandro; Pighin, Daniele; Basili, Roberto.

In: Computational Linguistics, Vol. 34, No. 2, 01.06.2008, p. 193-224.

Research output: Contribution to journalArticle

Moschitti, Alessandro ; Pighin, Daniele ; Basili, Roberto. / Tree kernels for semantic role labeling. In: Computational Linguistics. 2008 ; Vol. 34, No. 2. pp. 193-224.
@article{1483a71cd5a94641a9adc4612c83fc9c,
title = "Tree kernels for semantic role labeling",
abstract = "The availability of large scale data sets of manually annotated predicate-argument structures has recently favored the use of machine learning approaches to the design of automated semantic role labeling (SRL) systems. The main research in this area relates to the design choices for feature representation and for effective decompositions of the task in different learning models. Regarding the former choice, structural properties of full syntactic parses are largely employed as they represent ways to encode different principles suggested by the linking theory between syntax and semantics. The latter choice relates to several learning schemes over global views of the parses. For example, re-ranking stages operating over alternative predicate-argument sequences of the same sentence have shown to be very effective. In this article, we propose several kernel functions to model parse tree properties in kernelbased machines, for example, perceptrons or support vector machines. In particular, we define different kinds of tree kernels as general approaches to feature engineering in SRL. Moreover, we extensively experiment with such kernels to investigate their contribution to individual stages of an SRL architecture both in isolation and in combination with other traditional manually coded features. The results for boundary recognition, classification, and re-ranking stages provide systematic evidence about the significant impact of tree kernels on the overall accuracy, especially when the amount of training data is small. As a conclusive result, tree kernels allow for a general and easily portable feature engineering method which is applicable to a large family of natural language processing tasks.",
author = "Alessandro Moschitti and Daniele Pighin and Roberto Basili",
year = "2008",
month = "6",
day = "1",
doi = "10.1162/coli.2008.34.2.193",
language = "English",
volume = "34",
pages = "193--224",
journal = "Computational Linguistics",
issn = "0891-2017",
publisher = "MIT Press Journals",
number = "2",

}

TY - JOUR

T1 - Tree kernels for semantic role labeling

AU - Moschitti, Alessandro

AU - Pighin, Daniele

AU - Basili, Roberto

PY - 2008/6/1

Y1 - 2008/6/1

N2 - The availability of large scale data sets of manually annotated predicate-argument structures has recently favored the use of machine learning approaches to the design of automated semantic role labeling (SRL) systems. The main research in this area relates to the design choices for feature representation and for effective decompositions of the task in different learning models. Regarding the former choice, structural properties of full syntactic parses are largely employed as they represent ways to encode different principles suggested by the linking theory between syntax and semantics. The latter choice relates to several learning schemes over global views of the parses. For example, re-ranking stages operating over alternative predicate-argument sequences of the same sentence have shown to be very effective. In this article, we propose several kernel functions to model parse tree properties in kernelbased machines, for example, perceptrons or support vector machines. In particular, we define different kinds of tree kernels as general approaches to feature engineering in SRL. Moreover, we extensively experiment with such kernels to investigate their contribution to individual stages of an SRL architecture both in isolation and in combination with other traditional manually coded features. The results for boundary recognition, classification, and re-ranking stages provide systematic evidence about the significant impact of tree kernels on the overall accuracy, especially when the amount of training data is small. As a conclusive result, tree kernels allow for a general and easily portable feature engineering method which is applicable to a large family of natural language processing tasks.

AB - The availability of large scale data sets of manually annotated predicate-argument structures has recently favored the use of machine learning approaches to the design of automated semantic role labeling (SRL) systems. The main research in this area relates to the design choices for feature representation and for effective decompositions of the task in different learning models. Regarding the former choice, structural properties of full syntactic parses are largely employed as they represent ways to encode different principles suggested by the linking theory between syntax and semantics. The latter choice relates to several learning schemes over global views of the parses. For example, re-ranking stages operating over alternative predicate-argument sequences of the same sentence have shown to be very effective. In this article, we propose several kernel functions to model parse tree properties in kernelbased machines, for example, perceptrons or support vector machines. In particular, we define different kinds of tree kernels as general approaches to feature engineering in SRL. Moreover, we extensively experiment with such kernels to investigate their contribution to individual stages of an SRL architecture both in isolation and in combination with other traditional manually coded features. The results for boundary recognition, classification, and re-ranking stages provide systematic evidence about the significant impact of tree kernels on the overall accuracy, especially when the amount of training data is small. As a conclusive result, tree kernels allow for a general and easily portable feature engineering method which is applicable to a large family of natural language processing tasks.

UR - http://www.scopus.com/inward/record.url?scp=45749138033&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=45749138033&partnerID=8YFLogxK

U2 - 10.1162/coli.2008.34.2.193

DO - 10.1162/coli.2008.34.2.193

M3 - Article

VL - 34

SP - 193

EP - 224

JO - Computational Linguistics

JF - Computational Linguistics

SN - 0891-2017

IS - 2

ER -