An effective experts mining technique in online discussion forums

Abstract

Online question and answer forums have proven to be of great success in knowledge sharing. In these communities, askers post questions and wait for others to answer based on their personal knowledge. This process may lead to misguidance as the provided information has very less authenticity. Due to increasing demand of such forums, it is vital to find experts. Therefore, the shared information will be accurate and authentic in reality. In this paper, FB-Rank technique is presented with an aim to find expertise level of users. For this purpose, content based novel features are used. In addition, H-index, I-index and G-index methods are also used in this research domain. Experiments have been performed on StackOverflow dataset to evaluate the performance of FB-Rank. Kendall's, Spearman's rank and Osim correlation techniques are used to evaluate the effectiveness of the FB-Rank in a comparative manner.