StackOverflow.com: Analysis of user profiles: Identifying experts in CQA networks

Abstract

StackOverflow.com (SO), is a CQA network or a community Q&A network, where users can contribute to post questions to other users on specific topics of interest related to coding. The community is completely free and maintained by moderators and those users who are deemed “reputable”, based on their contributions in the past to StackOverflow.com. SO is completely free to use and posting and answering question merely requires one to register (for free) on SO. SO was founded in September 2008 and has since been rapidly growing in strength and building up a massive database of questions and answers. It is very commonly used as a reference point for coders to access information about errors that they may be facing. The user base of SO is quite vast, with over 1.3 Million users in total, of which around 440000 had answered at least one question1 In the recent past, considerable research has been done on analysing the SO database. SO provides all the information about its posts, votes and users (anonymized) in the form of data dumps which are publicly available for analysis. Over the recent past, SO has also started to be used for purposes other than simply answering questions. As a offshoot of the fact that a number of subject matter experts are regular contributors to SO, various companies have started using SO to identify new recruits. SO has a rewards and incentives based system which incentivises its users to be valuable contributors to the community. With the rewards, an SO user gains “Reputation”. This shall be described in more detail in the section below. The challenge with SO now is to identify who these subject-matter experts are. Past research has indicated that not all users with high reputation scores are equally competent in their respective domains.2 It has been suspected that some users have been “Gaming” the rewards system in place to quickly accumulate reputation. Along with the tangible benefits of reputation (in terms of various privileges on SO), there are a number of intangible benefits as well. When users misappropriate reputation by gaming the system, the community as a whole suffers. Thus, study and identification of traits of such users is of immense consequence not only to recruiters from outside (who would like to know exactly who they are recruiting beforehand), but also for SO itself, as it tries to build more and more value for its users as a CQA community.