34 IT Pro July/August 2015 Published by the IEEE Computer Society 1520-9202/15/$31.00 © 2015 IEEE
Simon Fong, University of Macau, Macau SAR, China
Suzy Zhou, Mozat, Singapore
Luiz Moutinho, University of Glasgow, Scotland
Online community question answering (CQA) services have gained
popularity among users wanting to voluntarily exchange solutions
without a fee. This article proposes classification using text analytics to
predict how likely a posted question is to be answered.
A
s a community-based knowledge-cre-
ation service,
1
the community question an-
swering (CQA) website model has risen
in popularity over the past few years, es-
pecially for topics requiring expert knowledge. CQA
lets users easily post questions online for peers to
answer voluntarily. The volume of questions an-
swered on such sites is predicted to soon surpass
those answered by library reference services.
2
In
fact, CQA websites are perceived as more precise
and trustworthy than traditional search engines
3
because answers come from users with real experi-
ence who, in turn, are questioners themselves and
expect the same quality in answers as they provide.
CQA websites pertaining to computer program-
ming in particular are gaining momentum among
programmers and software developers.
4
IT tech-
nologies evolve so rapidly that experienced devel-
opers can’t be proficient in every aspect of IT. For
example, the popular CQA programming website
Stack Overflow (http://stackoverflow.com) lets
developers post programming or IT-related ques-
tions in the hope that experts can provide answers.
By promoting good-quality Q&A and eliminating
uninteresting questions,
5
CQA websites will likely
prosper as they attract serious users.
One concern, however, is the high number of
unanswered questions on such websites. Re-
searchers found that out of 76,251 questions on
Yahoo Answers, 10,424 (about 13.67 percent) re-
ceived no answers.
6
Similarly, 20 percent of ques-
tions in Quora went unanswered, even though
DATA ANALYTICS
Text Analytics for
Predicting Question
Acceptance Rates