Paper: Word Clustering Based on Un-LP Algorithm

ACL ID W14-4505
Title Word Clustering Based on Un-LP Algorithm
Venue AHA!-Workshop on Information Discovery in Text
Year 2014

Word clustering which generalizes specific features cluster words in the same syntactic or seman- tic categories into a group. It is an effective approach to reduce feature dimensionality and feature sparseness which are clearly useful for many NLP applications. This paper proposes an unsu- pervised label propagation algorithm (Un-LP) for word clustering which uses multi-exemplars to represent a cluster. Experiments on a synthetic 2D dataset show the strong ability of self- correcting of the proposed algorithm. Besides, the experimental results on 20NG demonstrate that our algorithm outperforms the conventional cluster algorithms.