Semi-automatic labeling of coreferent named entities: an experimental study
Résumé
In this paper, we investigate the problem of matching coreferent named entities extracted from text collections in a robust way: our long-term goal is to build similarity methods without (or with the minimum amount of) prior knowledge. In this framework, string similarity measures are the main tool at our disposal. Here we focus on the problem of evaluating such a task, especially in finding a methodology to label the data in a semi-automatic way.
Domaines
Traitement du texte et du documentOrigine | Fichiers produits par l'(les) auteur(s) |
---|
Loading...