A CLUSTERING TECHNIQUE FOR THE VIETNAMESE WORD CATEGORIZATION

Abstract

In natural language processing, part-of-speech (POS) tagging plays an important role, as its output is the input of many other tasks (syntax analysis, semantic analysis. . . ). One of the problems related to POS tagging is to define the POS set. This could be solved using unsupervised machine learning methods. This paper presents an application of the DBSCAN clustering algorithm to classify Vietnamese words from a large corpus. The features used to characterize each word are naturally defined by the context of that word in a sentence. We use a large corpus containing sentences automatically extracted from the online Nhan Dan newspaper.