Ander Barrena

Learning Text Representations for 500

2018

Named Entity Disambiguation algorithms typically learn a single model for all target entities. In this paper we present a word expert model and train separate deep learning models for each target entity string, yielding 500K …