An improved algorithm for semantic clustering
Gabriel Castillo, Gerardo Sierra, John McNaught · Proceedings of the 1st international symposium on Information and communication technologies · 2003
Sierra & McNaught (2000) proposed a clustering algorithm based on analogy. It takes as input pairs of definitions sharing the same term headword, in the same domain but from different sources, compares these definitions and identifies couples of words in the same extended semantic context. These couples are then grouped to yield semantically-related clusters. Here, we further evaluate the clustering algorithm, and develop an improved version whose results are also evaluated.