A Study on Influential Features for Predicting Best Answers in Community Question-Answering Forums

Author:

Zoratto Valeria1,Godoy Daniela23,Aranda Gabriela N.1

Affiliation:

1. Facultad de Informática, Universidad Nacional del Comahue (UNComa), Neuquén 8300, Buenos Aires, Argentina

2. Facultad de Ciencias Exactas, Universidad Nacional del Centro de la Provincia de Buenos Aires (UNCPBA), ISISTAN, Tandil 7000, Buenos Aires, Argentina

3. Instituto Superior de Ingeniería de Software Tandil, ISISTAN, UNCPBA-CONICET, Tandil 7000, Buenos Aires, Argentina

Abstract

The knowledge provided by user communities in question-answering (QA) forums is a highly valuable source of information for satisfying user information needs. However, finding the best answer for a posted question can be challenging. User-generated content in forums can be of unequal quality given the free nature of natural language and the varied levels of user expertise. Answers to a question posted in a forum are compiled in a discussion thread, concentrating also posterior activity such as comments and votes. There are usually multiple reasons why an answer successfully fulfills a certain information need and gets accepted as the best answer among a (possibly) high number of answers. In this work, we study the influence that different aspects of answers have on the prediction of the best answers in a QA forum. We collected the discussion threads of a real-world forum concerning computer programming, and we evaluated different features for representing the answers and the context in which they appear in a thread. Multiple classification models were used to compare the performance of the different features, finding that readability is one of the most important factors for detecting the best answers. The goal of this study is to shed some light on the reasons why answers are more likely to receive more votes and be selected as the best answer for a posted question. Such knowledge enables users to enhance their answers which leads, in turn, to an improvement in the overall quality of the content produced in a platform.

Funder

ANPCyT

GIISCo Research Group

Publisher

MDPI AG

Subject

Information Systems

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3