Meta structures in knowledge graphs

This paper investigates meta structures, schema-level graphs that abstract connectivity information among a set of entities in a knowledge graph. Meta structures are useful in a variety of knowledge discovery tasks ranging from relatedness explanation to data retrieval. We formalize the meta structure computation problem and devise efficient automata-based algorithms. We introduce a meta structure-based relevance measure, which can retrieve entities related to those in input. We implemented our machineries in a visual tool called MEKoNG. We report on an extensive experimental evaluation, which confirms the suitability of our proposal from both the efficiency and effectiveness point of view.