Engineering Knowledge Graph From Patent Database

Abstract We propose a large, scalable engineering knowledge graph, comprising sets of real-world “facts” as < entity, relationship, entity > triples that are found in the patent database. apply set rules based on syntactic and lexical properties claims document to extract facts. aggregate these facts within each integrate aggregated across database obtain an graph. Such graph is expected support inference, reasoning, recalling various tasks. The has greater size coverage comparison with previously used graphs semantic networks literature.