A New Method of Text Feature Weighting Based on Position and Inter-class DistributionReportar como inadecuado




A New Method of Text Feature Weighting Based on Position and Inter-class Distribution - Descarga este documento en PDF. Documentación en PDF para descargar gratis. Disponible también para leer online.

The high dimension of text feature is the main bottleneck in text categorization beneath the vector space model. Text feature reduction is the core technology for text categorization. As a regular feature selection method, the multi-information is less efficient in practice. This article proposes one kind of an improvement MI algorithm. Aimed at the difference of the feature’s distribution in the class and in the text position, we optimize the weighting way in order to greater use of the category information. The result of the test shows that this method is better than the ordinary MI.

KEYWORDS

multi-information; feature selection; text categorization, feature reduction.

Cite this paper







Autor: Haifeng Liu, Zeyan Wang, Xueren Zhang, Qi Chen

Fuente: http://www.scirp.org/



DESCARGAR PDF




Documentos relacionados