Knowledge Graph Curation from Text Via Ontologies

Huge-scale knowledge graphs have become widespread on the web and are useful for a plethora of tasks. Knowledge graphs which are powered by machine learning use Natural Language Processing to build an extensive view of nodes, edges and labels by semantic enrichment in order to automate tasks like data structuring, text analysis and merging of data models. This research study put forth a semantically-driven and knowledge-centric strategy, which integrates machine learning and semantic inference for curating knowledge graphs from text using ontologies. The proposed framework amalgamates knowledge from the sources: Wikidata, DBpedia and Mediawiki to instill heterogeneity. The proffered framework transcends the baseline models by yielding an accuracy value of 98.22Cumulative Gain (NDCG) value of 0.97.