Abstract:
The CQA(Community question answering) is a perfect platform for the people who frequently participates to get the desired information of their interests. But if we want to find the expertise kind of users with relevant and authentic answers is the key challenge within these communities. In case, if we have to trust on someone’s opinion who is not well known by the communities, it is mandatory to find the credibility of the user. Because there is a scarcity of specialist organizations, professionals must rely on other resources when seeking knowledge. A huge online community, such as a Facebook discussion group, may have millions of members with a vast knowledge base containing millions of text documents. But a poor or low quality answer shows unqualified users therefore a priority is to find expert users. Expert-finding systems now in use assess user expertise based on the content of produced papers or one’s social position. We developed an expert finding technique called Expert Rank in this study, which assesses a user’s knowledge based on the content of authored documents and the social link of the authors. To identify the importance of topic, we used Latent Dirichletian Allocation (LDA) model to evaluate textual documents. Then we identified the social importance of users by using Google’s PageRank algorithm which is used by Google to rank websites. The modified PageRank calculates the social importance of authors who answered on same topic at different places. The more the social importance of authors, the higher the chances of being an expert. After that, we combine these expert ranking techniques using the cascade combine strategy. We evaluate our proposed algorithm Expert Rank using a most popular online community platform Stack Overflow. The experiments show that the proposed algorithm performs the best operations and shows the best performance when both topic modeling and social link analysis are considered
