Ranking in Co-effecting Multi-object/Link Types Networks

Abstract

Research on link based object ranking attracts increasing attention these years, which also brings computer science research and business marketing brand-new concepts, opportunities as well as a great deal of challenges. With prosperity of web pages search engine and widely use of social networks, recent graph-theoretic ranking approaches have achieved remarkable successes although most of them are focus on homogeneous networks studying. Previous study on co-ranking methods tries to divide heterogeneous networks into multiple homogeneous sub-networks and ties between different sub-networks. This paper proposes an efficient topic biased ranking method for bringing order to co-effecting heterogeneous networks among authors, papers and accepted institutions (journals/conferences) within one single random surfer. This new method aims to update ranks for different types of objects (author, paper, journals/conferences) at each random walk.