Paper: Learning Entity Representation for Entity Disambiguation

ACL ID P13-2006
Title Learning Entity Representation for Entity Disambiguation
Venue Annual Meeting of the Association of Computational Linguistics
Session Short Paper
Year 2013
Authors

We propose a novel entity disambigua- tion model, based on Deep Neural Net- work (DNN). Instead of utilizing simple similarity measures and their disjoint com- binations, our method directly optimizes document and entity representations for a given similarity measure. Stacked Denois- ing Auto-encoders are first employed to learn an initial document representation in an unsupervised pre-training stage. A su- pervised fine-tuning stage follows to opti- mize the representation towards the simi- larity measure. Experiment results show that our method achieves state-of-the-art performance on two public datasets with- out any manually designed features, even beating complex collective approaches.