An Empirical Study on Term Weights for Text Categorization

Abstract- In this paper, various term weighting methods for text categorization has been discussed. The terms represent the words, queries, phrases and indexing units and identify the texts. The supervised and unsupervised weighting methods to represent the prior information (supervised) or not(unsupervised) in the membership of training documents of categories were discussed. The review of various term weights approach under the text-based information processing presented will provide the necessary information for the researchers. This research is to provide a useful approach on the relationship among various term weight methods as well as to exploit the research domain. Keywords-Term weight, Text categorization, Information retrieval, Discriminative terms