Keyword search interface for path queries on ontology

Abstract

Today’s semantic web has a growing wealth of machine understandable metadata represented using markup languages like RDF, XML or OWL. There exists a plethora of query languages that aid is searching such data models. However, most real world searches involve queries expressed in natural language as it allows the user to get information without using complex formal query languages. This paper presents a search interface for path queries on ontologies, which accepts keywords and finds answers where each answer is a subgraph containing paths between nodes that match the keywords. Our approach for building such a system comprises of (1) a full-text search index for triples in the ontology (2) lexical and semantic query expansion to match user keywords to entities in the ontology, and (3) an algorithm which uses the Sparq2l path sequence indices to compute the answer subgraphs.