Expert Identification Using Topic Models in Semantic based Genealogical Graphs

Abstract

In scientometrics, semantically closer research articles tend to form a genealogical graph pattern which is used to derive explicit semantic lineage. The assumption of classic research article is that; the article has a high influence factor among the genealogy neighborhood. The candidates for identifying experts in each genealogical graph are chosen by finding one or more classic research articles from that graph and extracting the authors of those classic research articles. This paper proposes machine learning based approaches for mining the genealogical research paths which facilitate the inclusion of implicit citation/reference edges as well as indirectly linked citation lineage edges which are otherwise non-citing, to contribute effectively towards expert identification and ranking.