Content-Based Open Knowledge Graph Search: A Preliminary Study with OpenKG.CN

Abstract

Users rely on open data portals and search engines to find open knowledge graphs (KGs). However, existing systems only provide metadata-based KG search but ignore the contents of KGs, i.e., triples. In this paper, we present one of the first content-based search engines for open KGs. Our system CKGSE supports keyword-based KG search, KG snippet generation, KG profiling and browsing, all computed over KGs’ (large) contents rather than their (small) metadata. We implement a prototype with Chinese KGs crawled from OpenKG.CN and we report some preliminary results about the practicability of such a system.