Section: New Software and Platforms

KNEWS

Versatile Text-to-Knowledge Pipeline

Keyword: NLP

Functional Description

KNEWS is a versatile text-to-knowledge pipeline for machine reading, configurable to use different external modules for word sense disambiguation and entity linking. KNEWS works by running these components separately on a text, then it aligns the output of a semantic parser (Boxer) to the output of the other two modules, to produce complete semantic structures linked to DBpedia and Wordnet and represented as RDF graphs. KNEWS provides different kind of outputs: frame instances (based on the FrameBase scheme), word-aligned frames, and first-order logic formulas.

• Participants: Valerio Basile, Elena Cabrio and Fabien Gandon.

• Contact: Valerio Basile & Elena Cabrio

• URL: https://github.com/valeriobasile/learningbyreading