ENHANCED TFIDF ALGORITHM FOR TEXT CATEGORIZATION

Abstract

In this paper the enhanced features are used to fin distribution of a word in the document. The novel values assigned to a word are called features. These features like compactness of the appearances of the word and the position of the first appearance of the word. The proposed features are exploited by a tfidf style equation, and different features are combined using ensemble learning techniques. Experiments show that the distributional features are useful for text categorization. Text categorization is the task of assigning predefined categories to natural language text. With the widely used “bag-of-word†representation.