Paper: Jointly Identifying Entities and Extracting Relations in Encyclopedia Text via A Graphical Model Approach

ACL ID C10-2160
Title Jointly Identifying Entities and Extracting Relations in Encyclopedia Text via A Graphical Model Approach
Venue International Conference on Computational Linguistics
Session Poster Session
Year 2010
Authors

In this paper, we investigate the problem of en- tity identification and relation extraction from en- cyclopedia articles, and we propose a joint discrim- inative probabilistic model with arbitrary graphical structure to optimize all relevant subtasks simulta- neously. This modeling offers a natural formalism for exploiting rich dependencies and interactions between relevant subtasks to capture mutual ben- efits, as well as a great flexibility to incorporate a large collection of arbitrary, overlapping and non- independent features. We show the parameter es- timation algorithm of this model. Moreover, we propose a new inference method, namely collec- tive iterative classification (CIC), to find the most likely assignments for both entities and relations. We evaluate our model on real-world...