KSU KDD: Word Sense Induction by Clustering in Topic Space

Proceedings of the 5th International Workshop on Semantic Evaluation, ACL 2010, pages 367–370, Uppsala, Sweden, 15-16 July 2010. c⃝2010 Association for Computational Linguistics KSU KDD: Word Sense Induction by Clustering in Topic Space Wesam Elshamy, Doina Caragea, William H. Hsu Kansas State University {welshamy, dcaragea, bhsu}@ksu.edu Abstract We describe our language-independent un- supervised word sense induction system. This system only uses topic features to cluster different word senses in their global context topic space. Using unlabeled data, this system trains a latent Dirichlet allo- cation (LDA) topic model then uses it to infer the topics distribution of the test in- stances. By clustering these topics dis- tributions in their topic space we cluster them into different senses. Our hypothesis is that closeness in topic space reflects sim- ilarity between different word senses. This system participated in SemEval-2 word sense induction and disambiguation task and achieved the second highest V-measure score among all other systems. 1 Introduction Ambiguity of meaning is inherent in natural lan- guage because the deliverer of words tries to mini- mize the size of the vocabulary set he uses. There- fore, a sizable portion of this vocabulary is polyse- mous and the intended meaning of such words can be encoded in their context. Due to the knowledge acquisition bottleneck problem and scarcity in training data (Cai et al., 2007), unsupervised corpus based approaches could be favored over supervised ones in word sense disambiguation (WSD) tasks. Similar efforts in this area include work by Cai et al. (Cai et al., 2007) in which they use latent Dirichlet allocation (LDA) topic models to extract the global context topic and use it as a feature along other baseline features. Another technique uses clustering based approach with WordNet as an external resource for disambiguation without rely- ing on training data (Anaya-S´anchez et al., 2007). To disambiguate a polysemous word in a text document, we use the document topic distribution to represent its context. A document topic distri- bution is the probabilistic distribution of a docu- ment over a set of topics. The assumption is that: given two word senses and the topic distribution α β w z θ N M Figure 1: A graphical model for LDA of their context, the closeness between these two topic distributions in their topic space is an indi- cation of the similarity between those two senses. Our motivation behind building this system was the observation that the context of a polysemous word helps determining its sense to some degree. In our word sense induction (WSI) system, we use LDA to create a topic model for the given corpus and use it to infer the topic distribution of the documents containing the ambiguous words. This paper describes our WSI system which par- ticipated in SemEval-2 word sense induction and disambiguation task (Manandhar et al., 2010). 2 Latent Dirichlet allocation LDA is a probabilistic model for a collection of dis- crete data (Blei et al., 2003). It can be graphically represented as shown in Figure 1 as …

Content truncated to 500 words