A NEW EXPERT FINDING MODEL BASED ON TERM CORRELATION MATRIX
Ehsan Pournoor, Rezaee Noor Jalal · Journal of Information Processing and Management · 2015
Due to enormous volume of unstructured information available on the Web and inside organizations, finding an answer to the knowledge needed in a short time is difficult. For this reason, besides search engines which do not consider users individual characteristics, recommender systems were created which use user’s previous activities and other individual characteristics to help users find needed knowledge. Using recommender systems is increasing every day. By introducing expert people instead of recommending information to users, expert finder systems have provided this facility so that users ask their questions from experts. Having relation with experts not only causes information transition, but also transferring experiences and inceptions causes knowledge transition. In this paper we used university professors’ academic resume as expert people profile and then proposed a new expert finding model that recommends experts to users query. We used Term Correlation Matrix, Vector Space Model and PageRank algorithm and proposed a new hybrid model which outperforms conventional methods. This model can be used in internet environment, organizations and universities that experts have resume dataset.