Comparing Word Usage Using Latent Semantic Analysis

by
Simon Pai

Occasionally linguistics researchers have to analyze unfamiliar argots or sublanguages. One challenge is to recognize polysemous words (words that have multiple meanings) without knowing their meanings. Latent Semantic Analysis (LSA), a method that interprets words and documents in a given corpus as vectors, can be used to assign a degree of polysemy to each word. I will describe some experiments that use LSA to identify possible candidates for polysemous words.

Advised by Steve Taninoto

CSE 403
Wednesday
February 23, 2005
3:30 - 4:20 pm