基于朴素贝叶斯和支持向量机的短信智能分析系统设计

Development of an Intelligent SMS Analysis System Based on Naive Bayes and Support Vector Machine

为了解决垃圾短信的骚扰问题,提出了一种基于朴素贝叶斯和支持向量机的短信智能分析系统架构。考虑朴素贝叶斯的快速统计分类及支持向量机的增量训练等特点,将其应用于垃圾短信过滤中,并对关键词库进行及时更新,使得系统具有更好的自适应性。对某移动通信运营商提供的12万条短信进行反复实验,结果表明:该方法可有效地解决当前垃圾短信过滤系统中存在的问题。

In order to solve the harassment caused by spam messages, a spam message filtering system is proposed based on Native Bayes and support vector machine. This paper mainly utilizes Native Bayes advantage of rapid statistics classification and Support Vector Machine （SVM） incremental training characteristic in Spam Short Message filtering, and updates the keywords database in time in order to enhance the system＇s self-adaptability. Through repeated the experiment of 120,000 text messages from mobile operators, the experimental results show that this new method effectively deals with the above problems in the traditional spare short message filtering systems.