In this paper we analyze two question answering tasks: the TREC-8 question answering task and a set of reading comprehension exams. First, we show that Q/A systems perform better when there are multiple answer opportunities per question. Next, we analyze common approaches to two subproblems: term overlap for answer sentence identification, and answer typing for short answer extraction. We present general tools for analyzing the strengths and limitations of techniques for these subproblems. Our results quantify the limitations of both term overlap and answer typing to distinguish between competing answer candidates.
%0 Conference Paper
%1 BreckLightEtAl01ACL
%A Breck, Eric
%A Light, Marc
%A Mann, Gideon
%A Riloff, Ellen
%A Brown, Brianne
%A Anand, Pranav
%A Rooth, Mats
%A Thelen, Michael
%B Proceedings of the ACL 2001 Workshop on Open-Domain Question Answering, Toulouse, France
%D 2001
%K v1205 acl paper ai user interface language processing information retrieval answer test zzz.th.c4
%P 1-8
%R 10.3115/1117856.1117857
%T Looking Under the Hood: Tools for Diagnosing Your Question Answering Engine
%U http://www.aclweb.org/anthology/W01-1201
%X In this paper we analyze two question answering tasks: the TREC-8 question answering task and a set of reading comprehension exams. First, we show that Q/A systems perform better when there are multiple answer opportunities per question. Next, we analyze common approaches to two subproblems: term overlap for answer sentence identification, and answer typing for short answer extraction. We present general tools for analyzing the strengths and limitations of techniques for these subproblems. Our results quantify the limitations of both term overlap and answer typing to distinguish between competing answer candidates.
@inproceedings{BreckLightEtAl01ACL,
abstract = {In this paper we analyze two question answering tasks: the TREC-8 question answering task and a set of reading comprehension exams. First, we show that Q/A systems perform better when there are multiple answer opportunities per question. Next, we analyze common approaches to two subproblems: term overlap for answer sentence identification, and answer typing for short answer extraction. We present general tools for analyzing the strengths and limitations of techniques for these subproblems. Our results quantify the limitations of both term overlap and answer typing to distinguish between competing answer candidates.},
added-at = {2012-05-30T10:43:30.000+0200},
author = {Breck, Eric and Light, Marc and Mann, Gideon and Riloff, Ellen and Brown, Brianne and Anand, Pranav and Rooth, Mats and Thelen, Michael},
biburl = {https://www.bibsonomy.org/bibtex/2e938893c6f9a643e6366300be08760d5/flint63},
booktitle = {Proceedings of the ACL 2001 Workshop on Open-Domain Question Answering, Toulouse, France},
doi = {10.3115/1117856.1117857},
file = {ACL Anthology:2000-04/BreckLightEtAl01ACL.pdf:PDF},
groups = {public},
interhash = {5900e2302dcb6bf3b568e38722dbefcf},
intrahash = {e938893c6f9a643e6366300be08760d5},
keywords = {v1205 acl paper ai user interface language processing information retrieval answer test zzz.th.c4},
pages = {1-8},
timestamp = {2018-04-16T12:21:55.000+0200},
title = {Looking Under the Hood: Tools for Diagnosing Your Question Answering Engine},
url = {http://www.aclweb.org/anthology/W01-1201},
username = {flint63},
year = 2001
}