Effective shared representations with multitask learning for community question answering

Daniele Bonadiman, Antonio Uva, Alessandro Moschitti

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

1 Citation (Scopus)

Abstract

An important asset of using Deep Neural Networks (DNNs) for text applications is their ability to automatically engineer features. Unfortunately, DNNs usually require a lot of training data, especially for high-level semantic tasks such as community Question Answering (cQA). In this paper, we tackle the problem of data scarcity by learning the target DNN together with two auxiliary tasks in a multitask learning setting. We exploit the strong semantic connection between selection of comments relevant to (i) new questions and (ii) forum questions. This enables a global representation for comments, new and previous questions. The experiments of our model on a SemEval challenge dataset for cQA show a 20% relative improvement over standard DNNs.

Original languageEnglish
Title of host publicationShort Papers
PublisherAssociation for Computational Linguistics (ACL)
Pages726-732
Number of pages7
Volume2
ISBN (Electronic)9781510838604
Publication statusPublished - 1 Jan 2017
Event15th Conference of the European Chapter of the Association for Computational Linguistics, EACL 2017 - Valencia, Spain
Duration: 3 Apr 20177 Apr 2017

Other

Other15th Conference of the European Chapter of the Association for Computational Linguistics, EACL 2017
CountrySpain
CityValencia
Period3/4/177/4/17

Fingerprint

neural network
learning
community
semantics
engineer
assets
Question Answering
Neural Networks
experiment
ability

ASJC Scopus subject areas

  • Linguistics and Language
  • Language and Linguistics

Cite this

Bonadiman, D., Uva, A., & Moschitti, A. (2017). Effective shared representations with multitask learning for community question answering. In Short Papers (Vol. 2, pp. 726-732). Association for Computational Linguistics (ACL).

Effective shared representations with multitask learning for community question answering. / Bonadiman, Daniele; Uva, Antonio; Moschitti, Alessandro.

Short Papers. Vol. 2 Association for Computational Linguistics (ACL), 2017. p. 726-732.

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

Bonadiman, D, Uva, A & Moschitti, A 2017, Effective shared representations with multitask learning for community question answering. in Short Papers. vol. 2, Association for Computational Linguistics (ACL), pp. 726-732, 15th Conference of the European Chapter of the Association for Computational Linguistics, EACL 2017, Valencia, Spain, 3/4/17.
Bonadiman D, Uva A, Moschitti A. Effective shared representations with multitask learning for community question answering. In Short Papers. Vol. 2. Association for Computational Linguistics (ACL). 2017. p. 726-732
Bonadiman, Daniele ; Uva, Antonio ; Moschitti, Alessandro. / Effective shared representations with multitask learning for community question answering. Short Papers. Vol. 2 Association for Computational Linguistics (ACL), 2017. pp. 726-732
@inproceedings{8a40f222bbfd40999fb7ff705c29d719,
title = "Effective shared representations with multitask learning for community question answering",
abstract = "An important asset of using Deep Neural Networks (DNNs) for text applications is their ability to automatically engineer features. Unfortunately, DNNs usually require a lot of training data, especially for high-level semantic tasks such as community Question Answering (cQA). In this paper, we tackle the problem of data scarcity by learning the target DNN together with two auxiliary tasks in a multitask learning setting. We exploit the strong semantic connection between selection of comments relevant to (i) new questions and (ii) forum questions. This enables a global representation for comments, new and previous questions. The experiments of our model on a SemEval challenge dataset for cQA show a 20{\%} relative improvement over standard DNNs.",
author = "Daniele Bonadiman and Antonio Uva and Alessandro Moschitti",
year = "2017",
month = "1",
day = "1",
language = "English",
volume = "2",
pages = "726--732",
booktitle = "Short Papers",
publisher = "Association for Computational Linguistics (ACL)",

}

TY - GEN

T1 - Effective shared representations with multitask learning for community question answering

AU - Bonadiman, Daniele

AU - Uva, Antonio

AU - Moschitti, Alessandro

PY - 2017/1/1

Y1 - 2017/1/1

N2 - An important asset of using Deep Neural Networks (DNNs) for text applications is their ability to automatically engineer features. Unfortunately, DNNs usually require a lot of training data, especially for high-level semantic tasks such as community Question Answering (cQA). In this paper, we tackle the problem of data scarcity by learning the target DNN together with two auxiliary tasks in a multitask learning setting. We exploit the strong semantic connection between selection of comments relevant to (i) new questions and (ii) forum questions. This enables a global representation for comments, new and previous questions. The experiments of our model on a SemEval challenge dataset for cQA show a 20% relative improvement over standard DNNs.

AB - An important asset of using Deep Neural Networks (DNNs) for text applications is their ability to automatically engineer features. Unfortunately, DNNs usually require a lot of training data, especially for high-level semantic tasks such as community Question Answering (cQA). In this paper, we tackle the problem of data scarcity by learning the target DNN together with two auxiliary tasks in a multitask learning setting. We exploit the strong semantic connection between selection of comments relevant to (i) new questions and (ii) forum questions. This enables a global representation for comments, new and previous questions. The experiments of our model on a SemEval challenge dataset for cQA show a 20% relative improvement over standard DNNs.

UR - http://www.scopus.com/inward/record.url?scp=85021752447&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85021752447&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:85021752447

VL - 2

SP - 726

EP - 732

BT - Short Papers

PB - Association for Computational Linguistics (ACL)

ER -