Discriminating Among Word Senses Using McQuitty's Similarity Analysis

This paper presents an unsupervised method for discriminating among the senses of a given target word based on the context in which it occurs. Instances of a word that occur in similar contexts are grouped together via McQuitty’s Similarity Analysis, an agglomerative clustering algorithm. The context in which a target word occurs is represented by surface lexical features such as unigrams, bigrams, and second order co-occurrences. This paper summarizes our approach, and describes the results of a preliminary evaluation we have carried out using data from the SENSEVAL-2 English lexical sample and the line corpus.

• Type: Conference
• Year: 2003
• Where: NAACL
• Authors: Amruta Purandare
• Keywords: Agglomerative Clustering Algorithm, NAACL 2003, NAACL 2007, Surface Lexical Features, Target Word