Synapse at CAp 2017 NER challenge: Fasttext CRF

Damien Sileo (1, 2) , Camille Pradel (1) , Philippe Muller (2) , Tim van de Cruys (2)

1 Synapse Développement
2 IRIT-MELODI - MEthodes et ingénierie des Langues, des Ontologies et du DIscours

We present our system for the CAp 2017 NER challenge which is about named entity recognition on French tweets. Our system leverages unsupervised learning on a larger dataset of French tweets to learn features feeding a CRF model. It was ranked first without using any gazetteer or structured external data, with an F-measure of 58.89\%. To the best of our knowledge, it is the first system to use fasttext embeddings (which include subword representations) and an embedding-based sentence representation for NER.

Mots clés: Named entity recognition; CRF; Fasttext; Unsupervised learning; Word vectors

Domaines: Informatique et langage [cs.CL]

Damien Sileo, Camille Pradel, Philippe Muller, Tim van de Cruys. Synapse at CAp 2017 NER challenge: Fasttext CRF. 19e Conference francophone sur l'Apprentissage Automatique (CAp 2017), Jun 2017, Grenoble, France. pp.1-5. ⟨hal-02879724⟩