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

Abstract

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.We present ANCOR, a French corpus annotated with coreference relations which is freely available and large enough to serve the needs of data-driven approaches

Referring expressions and coreference chains in French: the Democrat project and its exploration of links between textual linguistics and corpus linguistics

“…Contrairement à la première approche, la seconde nécessite de disposer d'un corpus, qui sert de corpus d'apprentissage. Un premier ensemble de techniques a été exploré avec le système CROC, « Coreference Resolution for Oral Corpus », développé par des membres du projet Democrat juste avant le lancement de celui-ci (DÉSOYER et al 2018). Ce système présentait cependant une limitation forte : il nécessitait de disposer d'un texte déjà annoté en expressions référentielles.…”