Review of Dimension Reduction Technologies in Text Categorization for Military Intelligence

To ensure the quality and speed for military decision-making and improve capability of military intelligence text analysis,an effective text categorization becomes a very important technology.The paper describes the main difficulties and problems in the text categorization,including super sparse text space and curse of dimensionality.Feature selection technologies on how to search optimal or suboptimal feature subnet and feature extraction technologies,such as PCA,FDA and LSI,are introduced in detail.Existing technological problems and future development are given.