Désambiguïsation d’entités nommées par apprentissage de modèles d’entités à large échelle

Hani Daher (1, 2) , Romaric Besançon (1) , Olivier Ferret (1) , Hervé Le Borgne (1) , Anne-Laure Daquo (1) , Youssef Tamaazousti (1, 3)

1 DIASI - Département Intelligence Ambiante et Systèmes Interactifs
2 LVIC - Laboratoire Vision et Ingénierie des Contenus
3 MICS - Mathématiques et Informatique pour la Complexité et les Systèmes

The objective of Entity Linking is to connect an entity mention in a text to a known entity in a knowledge base. The general approach for this task is to generate, for a given mention, a set of candidate entities from the base and determine, in a second step, the best one. This paper focuses on this last step and proposes a method based on learning a function that discriminates an entity from its most ambiguous ones. We adopt a model that is able to deal with large knowledge bases. Thus our contribution lies in the strategy to learn efficiently such a model. We propose three strategies with different efficiency/performance tradeoff. The approach is experimentally validated on six datasets of the TAC evaluation campaigns by using Freebase and DBpedia as reference knowledge bases

Keywords: Entity IR; Entity retrieval; Entity linking; Negative sample selection

Domains: Information Retrieval [cs.IR]; Document and Text Processing

Hani Daher, Romaric Besançon, Olivier Ferret, Hervé Le Borgne, Anne-Laure Daquo, et al.. Désambiguïsation d’entités nommées par apprentissage de modèles d’entités à large échelle. 14ème COnférence en Recherche d'Information et Applications (CORIA 2017), 2017, Marseille, France. pp.185-200. ⟨cea-01857881⟩