LSTC System for Chinese Word Sense Induction

This paper presents the Chinese word sense Induction system of Leshan Teachers’ College. The system participates in the Chinese word sense Induction of task 4 in Back offs organized by the Chinese Information Processing Society of China (CIPS) and SIGHAN. The system extracts neighbor words and their POSs centered in the target words and selected the best one of four cluster algorithms: Simple KMeans, EM, Farthest First and Hierarchical Cluster based on training data. We obtained the F-Score of 60.5% on the training data otherwise the F-Score is 57.89% on the test data provided by organizers.