Predicting the quality of questions on Stackoverflow

Proceedings of Recent Advances in Natural Language Processing, pages 32–40, Hissar, Bulgaria, Sep 7–9 2015. Predicting the Quality of Questions on Stackoverflow Antoaneta Baltadzhieva Tilburg University a baltadzhieva@yahoo.de Grzegorz Chrupała Tilburg University g.a.chrupala@uvt.nl Abstract Community Question Answering websites (CQA) have a growing popularity as a way of providing and searching of information. CQA attract users as they provide a direct and rapid way to find the desired informa- tion. As recognizing good questions can improve the CQA services and the user’s experience, the current study focuses on question quality instead. Specifically, we predict question quality and investigate the features which influence it. The influence of the question tags, length of the question title and body, presence of a code snippet, the user reputation and terms used to for- mulate the question are tested. For each set of dependent variables, Ridge regres- sion models are estimated. The results in- dicate that the inclusion of terms in the models improves their predictive power. Additionally, we investigate which lexi- cal terms determine high and low qual- ity questions. The terms with the high- est and lowest coefficients are semanti- cally analyzed. The analysis shows that terms predicting high quality are terms ex- pressing, among others, excitement, nega- tive experience or terms regarding excep- tions. Terms predicting low quality ques- tions are terms containing spelling errors or indicating off-topic questions and inter- jections. 1 Introduction CQA websites provide an interface for users to exchange and share knowledge. The user ask- ing a question lacks knowledge of a specific topic and searches for an expert to provide the desired knowledge. In this way, the asker is querying a topic and the experts are the source of informa- tion, replacing other sources like documents or databases. However, the search results may not provide an exact solution to the user’s problem. Although the idea of receiving a direct response to an information need sounds very appealing, CQA websites also involve risk as the quality of the pro- vided information is not guaranteed. An important difference between user-generated content and tra- ditional content is the range of the content qual- ity: user-generated content shows a higher vari- ance in quality (Agichtein et al., 2008) than tradi- tional content (Anderson, 2006). Stack Overflow (SO) is a CQA website in the field of computer programming. Access is free and answers are voted according to the asker’s sat- isfaction1. The asker can tag a question to indicate a specific subject. Users can vote questions, an- swers and edits to indicate how helpful they were. The votes determine the user’s reputation. In or- der to create a high-quality library of questions and their answers, SO allows users not only to post questions or answers but also to edit them. Despite the encouragement of SO and the of- fered opportunities to maintain the content qual- ity, a lot of questions on SO are not answered. With the increase in popularity of SO, not only the number of questions and the number of new …

Content truncated to 500 words