Scaling up knowledge graph creation to large and heterogeneous data sources

RDF knowledge graphs (KG) are powerful data structures to represent factual statements created from heterogeneous sources. KG creation is laborious and demands management techniques be executed efficiently. This paper tackles the problem of automatic generation processes declaratively specified; it proposes for planning transforming into triples following mapping assertions specified in Mapping Language (RML). Given a set assertions, planner provides an optimized execution plan by partitioning scheduling assertions. First, assesses number partitions considering sources, type associations between different After providing list that belong each partition, determines their order. A greedy algorithm implemented generate partitions’ bushy tree plan. Bushy plans translated operating system commands guide order indicated tree. The proposed optimization approach evaluated over state-of-the-art RML-compliant engines, existing benchmarks sources RML maps. Our experimental results suggest performance studied engines can considerably improved, particularly complex setting with numerous maps large As result, time out cases enabled produce at least portion applying planner.