Relationship search over knowledge graphs

Abstract

The recent explosive growth of knowledge graphs (KGs) has enhanced a wide range of Web applications. Among others, the established task of relationship search benefits from KGs since complex relationships that were hidden and distributed in multiple webpages are now explicitly available as subgraphs of an integrated KG. However, new challenges arose as to the efficiency of search over large KGs. In this article, we introduce our recent studies on relationship search, including search algorithms based on novel index structures, methods for relationship clustering to support result exploration, and query relaxation techniques to provide alternative results for failed searches. We also present emerging applications and discuss future research directions.