Design And Implementation Of Text Classification System

With the coming of the information age, the Internet develops rapidly. The ways of information communication that emphasized interaction, convenience, and real time has gradually become the mainstream mass communication channels. As a result, a large number of text messages containing a wealth of information, such as entertainment, agriculture, military, economic and many other areas, hold great value waiting to dig. So it has become our urgent study to analyze large-scale texts in a more effective and accurate way. The appearance of text categorization method helps people filter spam and analyze the useful information in a more effective and faster way, which is benefit in all areas of our life.The study of this paper is establishing a text classification system for the project. To build a common and extensible text classification system, we embed the classic feature selection methods, self-designed feature selection method and classic classification algorithms into the system. This system could help the researchers avoid repetitive coding and save the resources. The main work of this paper is as follows :(1)Build an extensible, friendly interactive and visualization text classification system. Introduce the major functions of the system and verify the effectiveness of improved algorithm by testing.(2)Introduce the main process of text classification and focus on the contrastive analysis of feature selection methods and classification algorithms.(3) A new feature selection method, Feature Selection Based on Improved Particle Swarm Optimization Algorithm, has been presented in this paper in order to enrich the algorithms embedded in the system. This new method could avoid premature convergence and improve classification performance by the particle initial perturbations and improving standard PSO inertia weight to non-liner form.