Recommending Experts in Q&A Communities by Weighted HITS Algorithm

Recently Q&A portals, such as Yahoo! Answers etc., are rapidly growing in popularity. Recommending experts is the crucial problem for the development of Q&A communities. In this paper, we focus on the problem of measuring the reputation of users, in which a question is given by an asker and system returns a ranked list of top K experts. The proposed approach consists of two steps: (1) analysis network structure of Q&A; (2) describe the weighted HITS algorithm, calculate reputation score of each user and recommend experts for asker who post new question. Experiments are carried out on a real-world data crawled from Yahoo! Answers during Mar 24 2008 to Oct 13 2008, which consists of 94759 questions, 473481 answers and 73511 users. The experimental results reveal significant improvements over the baseline methods.