A Semantic-Rich Similarity Measure in Heterogeneous Information Networks

摘要

Measuring the similarities between objects in information networks hasfundamental importance in recommendation systems, clustering and web search.The existing metrics depend on the meta path or meta structure specified byusers. In this paper, we propose a stratified meta structure based similarity$SMSS$ in heterogeneous information networks. The stratified meta structure canbe constructed automatically and capture rich semantics. Then, we define thecommuting matrix of the stratified meta structure by virtue of the commutingmatrices of meta paths and meta structures. As a result, $SMSS$ is defined byvirtue of these commuting matrices. Experimental evaluations show that theproposed $SMSS$ on the whole outperforms the state-of-the-art metrics in termsof ranking and clustering.