String Vector based AHC Algorithm for Word Clustering from News Articles

摘要

In this research, we propose the AHC version where words are encoded into string vectors, instead of numerical vectors, as the approach to the word clustering. In the previous works, the better text categorization performances from encoding texts into string vectors than into numerical vectors are shown and we need to reinforce both the word clustering and the text clustering by connecting them with each other. In this research, words are encoded into string vectors each of which consists of text identifiers, the semantic similarity between two string vectors is defined as the operation on string vectors, and the AHC algorithm is modified by adopting the proposed similarity metric. We adopt the clustering index as the evaluation metric, and validate empirically that the proposed AHC version is better than the traditional version. In future, we connect mutually the word clustering with the text clustering for reinforcing them at same time.