Semantic role labeling for knowledge graph extraction from text

Abstract This paper introduces , a new semantic role labeling method that transforms text into frame-oriented knowledge graph. It performs dependency parsing, identifies the words evoke lexical frames, locates roles and fillers for each frame, runs coercion techniques, formalizes results as formal representation complies with frame semantics used in Framester, factual-linguistic linked data resource. We tested our on WSJ section of Peen Treebank annotated VerbNet PropBank labels Brown corpus. The evaluation has been performed according to CoNLL Shared Task Joint Parsing Syntactic Semantic Dependencies. obtained precision, recall, F1 values indicate TakeFive is competitive other existing methods such SEMAFOR, Pikes, PathLSTM, FRED. finally discuss how combine FRED, obtaining higher measure.