A Social Trust Metric For Scholarly Reputation Mining

Abstract

Trust has been described as an intrinsic component of any social relation. Trust mainly refers to a measure of confidence on an entity that would behave in an expected manner. Academic social networking sites enable researchers to communicate, and share publications. This paper aims to rank both researchers and their productivity in terms of their scientific paper they are publishing. A trust model is proposed that utilizes the metadata of researchers and their papers, extracted from academic social networks, in order to produce two trust values, one for a researcher and another for a scientific paper. The utilized metadata for the researchers includes (total publication, total work citation, followers, h-index). Propagation of trust score using top co-authors is also considered for authors. The metadata of papers are: calculated author score, paper citation, and references citation. Each of those individual factors is assigned a specific weight based on user preference and AHP ranking method is applied. Individual metadata are aggregated to a collective one by considering aggregation weights for each feature and applying AHP ranking method. Experimental show that the proposed model provide a high accuracy value when compared using ground truth data from Google Scholar and global H-index.