An analogy-based method for semantic clustering
Gerardo Sierra, John McNaught · Research Explorer (The University of Manchester) · 2000
An analogy-based clustering method is proposed, through the alignment of definitions from two different sources. The method relies on the assumption that two authors use different words to express a definition. The algorithm introduced here is analogy-based, and starts from calculating the Levenshtein distance, which is a variation of the edit distance, and allows us to align the definitions. As a measure of similarity, the concept of longest collocation couple is introduced, which is the basis of clustering similar words. The process iterates, replacing similar pairs of words in the definitions until no new clusters are found.