Simple similarity-based question answering strategies for biomedical text

David Martinez, Andrew MacKinlay, Diego Molla-Aliod, Lawrence Cavedon, Karin Verspoor

Research output: Contribution to journalConference paperResearch

Abstract

We introduce an approach to question answering in the biomedical domain that utilises similarity matching of question/answer pairs in a document, or a set of background documents, to select the best answer to a multiple-choice question. We explored a range of possible similarity matching methods, ranging from simple word overlap, to dependency graph matching, to feature-based vector similarity models that incorporate lexical, syntactic and/or semantic features. We found that while these methods performed reasonably well on a small training set, they did not generalise well to the final test data.

LanguageEnglish
Pages1-13
Number of pages13
JournalCEUR Workshop Proceedings
Volume1178
Publication statusPublished - 2012

Fingerprint

Syntactics
Semantics

Cite this

Martinez, D., MacKinlay, A., Molla-Aliod, D., Cavedon, L., & Verspoor, K. (2012). Simple similarity-based question answering strategies for biomedical text. CEUR Workshop Proceedings, 1178, 1-13.
Martinez, David ; MacKinlay, Andrew ; Molla-Aliod, Diego ; Cavedon, Lawrence ; Verspoor, Karin. / Simple similarity-based question answering strategies for biomedical text. In: CEUR Workshop Proceedings. 2012 ; Vol. 1178. pp. 1-13.
@article{3994fa8921c04640b8d41044c5c409c7,
title = "Simple similarity-based question answering strategies for biomedical text",
abstract = "We introduce an approach to question answering in the biomedical domain that utilises similarity matching of question/answer pairs in a document, or a set of background documents, to select the best answer to a multiple-choice question. We explored a range of possible similarity matching methods, ranging from simple word overlap, to dependency graph matching, to feature-based vector similarity models that incorporate lexical, syntactic and/or semantic features. We found that while these methods performed reasonably well on a small training set, they did not generalise well to the final test data.",
author = "David Martinez and Andrew MacKinlay and Diego Molla-Aliod and Lawrence Cavedon and Karin Verspoor",
year = "2012",
language = "English",
volume = "1178",
pages = "1--13",
journal = "CEUR Workshop Proceedings",
issn = "1613-0073",
publisher = "RWTH Aachen University",

}

Martinez, D, MacKinlay, A, Molla-Aliod, D, Cavedon, L & Verspoor, K 2012, 'Simple similarity-based question answering strategies for biomedical text', CEUR Workshop Proceedings, vol. 1178, pp. 1-13.

Simple similarity-based question answering strategies for biomedical text. / Martinez, David; MacKinlay, Andrew; Molla-Aliod, Diego; Cavedon, Lawrence; Verspoor, Karin.

In: CEUR Workshop Proceedings, Vol. 1178, 2012, p. 1-13.

Research output: Contribution to journalConference paperResearch

TY - JOUR

T1 - Simple similarity-based question answering strategies for biomedical text

AU - Martinez, David

AU - MacKinlay, Andrew

AU - Molla-Aliod, Diego

AU - Cavedon, Lawrence

AU - Verspoor, Karin

PY - 2012

Y1 - 2012

N2 - We introduce an approach to question answering in the biomedical domain that utilises similarity matching of question/answer pairs in a document, or a set of background documents, to select the best answer to a multiple-choice question. We explored a range of possible similarity matching methods, ranging from simple word overlap, to dependency graph matching, to feature-based vector similarity models that incorporate lexical, syntactic and/or semantic features. We found that while these methods performed reasonably well on a small training set, they did not generalise well to the final test data.

AB - We introduce an approach to question answering in the biomedical domain that utilises similarity matching of question/answer pairs in a document, or a set of background documents, to select the best answer to a multiple-choice question. We explored a range of possible similarity matching methods, ranging from simple word overlap, to dependency graph matching, to feature-based vector similarity models that incorporate lexical, syntactic and/or semantic features. We found that while these methods performed reasonably well on a small training set, they did not generalise well to the final test data.

UR - http://www.scopus.com/inward/record.url?scp=84922021520&partnerID=8YFLogxK

M3 - Conference paper

VL - 1178

SP - 1

EP - 13

JO - CEUR Workshop Proceedings

T2 - CEUR Workshop Proceedings

JF - CEUR Workshop Proceedings

SN - 1613-0073

ER -