Publication Type

Conference Proceeding Article

Publication Date

12-2016

Abstract

Relation classification is the task of classifying the semantic relations between entity pairs intext. Observing that existing work has not fully explored using different representations forrelation instances, especially in order to better handle the asymmetry of relation types, in thispaper, we propose a neural network based method for relation classification that combines theraw sequence and the shortest dependency path representations of relation instances and usesmirror instances to perform pairwise relation classification. We evaluate our proposed modelson two widely used datasets: SemEval-2010 Task 8 and ACE-2005. The empirical results showthat our combined model together with mirror instances achieves the state-of-the-art results onboth datasets.

Discipline

Databases and Information Systems | Systems Architecture

Research Areas

Data Management and Analytics

Publication

Proceedings of COLING 2016: The 26th International Conference on Computational Linguistics: Technical Papers, Japan, 2016 December 11-17

First Page

2373

Last Page

2377

ISBN

9784879747020

Publisher

Creative Commons

City or Country

Osaka, Japan

Creative Commons License

Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.

Additional URL

http://coling2016.anlp.jp/doc/main.pdf

Download

Share

COinS