Online latent structure training for language acquisition

Michael Connor, Cynthia L Fisher, Dan Roth

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

Abstract

A fundamental step in sentence comprehension involves assigning semantic roles to sentence constituents. To accomplish this, the listener must parse the sentence, find constituents that are candidate arguments, and assign semantic roles to those constituents. Where do children learning their first languages begin in solving this problem? Even assuming children can derive a rough meaning for the sentence from the situation, how do they begin to map this meaning to the structure and the structure to the form of the sentence? In this paper we use feedback from a semantic role labeling (SRL) task to improve the intermediate syntactic representations that feed the SRL. We accomplish this by training an intermediate classifier using signals derived from latent structure optimization techniques. By using a separate classifier to predict internal structure we see benefits due to knowledge embedded in the classifier's feature representation. This extra structure allows the system to begin to learn using weaker, more plausible semantic feedback.

Original languageEnglish (US)
Title of host publicationIJCAI 2011 - 22nd International Joint Conference on Artificial Intelligence
Pages1782-1787
Number of pages6
DOIs
StatePublished - Dec 1 2011
Event22nd International Joint Conference on Artificial Intelligence, IJCAI 2011 - Barcelona, Catalonia, Spain
Duration: Jul 16 2011Jul 22 2011

Publication series

NameIJCAI International Joint Conference on Artificial Intelligence
ISSN (Print)1045-0823

Other

Other22nd International Joint Conference on Artificial Intelligence, IJCAI 2011
CountrySpain
CityBarcelona, Catalonia
Period7/16/117/22/11

Fingerprint

Semantics
Classifiers
Labeling
Feedback
Syntactics

ASJC Scopus subject areas

  • Artificial Intelligence

Cite this

Connor, M., Fisher, C. L., & Roth, D. (2011). Online latent structure training for language acquisition. In IJCAI 2011 - 22nd International Joint Conference on Artificial Intelligence (pp. 1782-1787). (IJCAI International Joint Conference on Artificial Intelligence). https://doi.org/10.5591/978-1-57735-516-8/IJCAI11-299

Online latent structure training for language acquisition. / Connor, Michael; Fisher, Cynthia L; Roth, Dan.

IJCAI 2011 - 22nd International Joint Conference on Artificial Intelligence. 2011. p. 1782-1787 (IJCAI International Joint Conference on Artificial Intelligence).

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

Connor, M, Fisher, CL & Roth, D 2011, Online latent structure training for language acquisition. in IJCAI 2011 - 22nd International Joint Conference on Artificial Intelligence. IJCAI International Joint Conference on Artificial Intelligence, pp. 1782-1787, 22nd International Joint Conference on Artificial Intelligence, IJCAI 2011, Barcelona, Catalonia, Spain, 7/16/11. https://doi.org/10.5591/978-1-57735-516-8/IJCAI11-299
Connor M, Fisher CL, Roth D. Online latent structure training for language acquisition. In IJCAI 2011 - 22nd International Joint Conference on Artificial Intelligence. 2011. p. 1782-1787. (IJCAI International Joint Conference on Artificial Intelligence). https://doi.org/10.5591/978-1-57735-516-8/IJCAI11-299
Connor, Michael ; Fisher, Cynthia L ; Roth, Dan. / Online latent structure training for language acquisition. IJCAI 2011 - 22nd International Joint Conference on Artificial Intelligence. 2011. pp. 1782-1787 (IJCAI International Joint Conference on Artificial Intelligence).
@inproceedings{4e2099ffba794a17a4e26363f46c1094,
title = "Online latent structure training for language acquisition",
abstract = "A fundamental step in sentence comprehension involves assigning semantic roles to sentence constituents. To accomplish this, the listener must parse the sentence, find constituents that are candidate arguments, and assign semantic roles to those constituents. Where do children learning their first languages begin in solving this problem? Even assuming children can derive a rough meaning for the sentence from the situation, how do they begin to map this meaning to the structure and the structure to the form of the sentence? In this paper we use feedback from a semantic role labeling (SRL) task to improve the intermediate syntactic representations that feed the SRL. We accomplish this by training an intermediate classifier using signals derived from latent structure optimization techniques. By using a separate classifier to predict internal structure we see benefits due to knowledge embedded in the classifier's feature representation. This extra structure allows the system to begin to learn using weaker, more plausible semantic feedback.",
author = "Michael Connor and Fisher, {Cynthia L} and Dan Roth",
year = "2011",
month = "12",
day = "1",
doi = "10.5591/978-1-57735-516-8/IJCAI11-299",
language = "English (US)",
isbn = "9781577355120",
series = "IJCAI International Joint Conference on Artificial Intelligence",
pages = "1782--1787",
booktitle = "IJCAI 2011 - 22nd International Joint Conference on Artificial Intelligence",

}

TY - GEN

T1 - Online latent structure training for language acquisition

AU - Connor, Michael

AU - Fisher, Cynthia L

AU - Roth, Dan

PY - 2011/12/1

Y1 - 2011/12/1

N2 - A fundamental step in sentence comprehension involves assigning semantic roles to sentence constituents. To accomplish this, the listener must parse the sentence, find constituents that are candidate arguments, and assign semantic roles to those constituents. Where do children learning their first languages begin in solving this problem? Even assuming children can derive a rough meaning for the sentence from the situation, how do they begin to map this meaning to the structure and the structure to the form of the sentence? In this paper we use feedback from a semantic role labeling (SRL) task to improve the intermediate syntactic representations that feed the SRL. We accomplish this by training an intermediate classifier using signals derived from latent structure optimization techniques. By using a separate classifier to predict internal structure we see benefits due to knowledge embedded in the classifier's feature representation. This extra structure allows the system to begin to learn using weaker, more plausible semantic feedback.

AB - A fundamental step in sentence comprehension involves assigning semantic roles to sentence constituents. To accomplish this, the listener must parse the sentence, find constituents that are candidate arguments, and assign semantic roles to those constituents. Where do children learning their first languages begin in solving this problem? Even assuming children can derive a rough meaning for the sentence from the situation, how do they begin to map this meaning to the structure and the structure to the form of the sentence? In this paper we use feedback from a semantic role labeling (SRL) task to improve the intermediate syntactic representations that feed the SRL. We accomplish this by training an intermediate classifier using signals derived from latent structure optimization techniques. By using a separate classifier to predict internal structure we see benefits due to knowledge embedded in the classifier's feature representation. This extra structure allows the system to begin to learn using weaker, more plausible semantic feedback.

UR - http://www.scopus.com/inward/record.url?scp=84881048672&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84881048672&partnerID=8YFLogxK

U2 - 10.5591/978-1-57735-516-8/IJCAI11-299

DO - 10.5591/978-1-57735-516-8/IJCAI11-299

M3 - Conference contribution

AN - SCOPUS:84881048672

SN - 9781577355120

T3 - IJCAI International Joint Conference on Artificial Intelligence

SP - 1782

EP - 1787

BT - IJCAI 2011 - 22nd International Joint Conference on Artificial Intelligence

ER -