Experts Ranking in Online Communities Using Combination of Text and Link Analysis

Abstract

Online communities are a question answering environment where individuals can express their opinions freely. Quality of the
information generated in the online community is dependent on the individual expertise level. If the individual has higher level of expertise,
shared knowledge in online community is valuable and reliable. Experts ranking methods used for determining expertise level of individuals
and evaluating the accuracy of shared knowledge. In this study, a novel hybrid method for ranking experts in online communities is
presented. This approach incorporates users' relationship and content of users' answers to finding expert users in an online community. This
method is applicable to all online communities and only corpus in the field of online community is needed to accomplish that. We evaluated
our proposed method on Java online community and Cryptography section of StackExchange online community. Correlation between scores
of our method and scores of expert users introduced in both online communities exceeds 0.8, which is highly a reasonable value.