Coreference Resolution for French Oral Data: Machine Learning Experiments with ANCOR

Adèle Désoyer (1, 2) , Frédéric Landragin (3, 4, 2) , Isabelle Tellier (2, 3, 4) , Anaïs Lefeuvre (5) , Jean-Yves Antoine (5) , Marco Dinarelli (2, 3, 4)

1 MoDyCo - Modèles, Dynamiques, Corpus
2 Lattice - Lattice - Langues, Textes, Traitements informatiques, Cognition - UMR 8094
3 PSL - Université Paris sciences et lettres
4 USPC - Université Sorbonne Paris Cité
5 BDTLN - Bases de données et traitement des langues naturelles

We present CROC (Coreference Resolution for Oral Corpus), the first machine learning system for coreference resolution in French. One specific aspect of the system is that it has been trained on data that come exclusively from transcribed speech, namely ANCOR (ANaphora and Coreference in ORal corpus), the first large-scale French corpus with anaphorical relation annotations. In its current state, the CROC system requires pre-annotated mentions. We detail the features used for the learning algorithms, and we present a set of experiments with these features. The scores we obtain are close to those of state-of-the-art systems for written English.

Mots clés: mention-pair model; dialogue corpus; coreference resolution; machine learning

Domaines: Informatique et langage [cs.CL]; Linguistique; Sciences de l'information et de la communication

Adèle Désoyer, Frédéric Landragin, Isabelle Tellier, Anaïs Lefeuvre, Jean-Yves Antoine, et al.. Coreference Resolution for French Oral Data: Machine Learning Experiments with ANCOR. 17th International Conference on Intelligent Text Processing and Computational Linguistics (CICLing'2016), Apr 2016, Konya, Turkey. ⟨hal-01344977⟩