一种用于微信信息分类的改进贝叶斯算法

An Optimized Bayesian Classification Algorithm for WeChat Information

微信的快速普及加快了信息的传播,随之而来的广告、诈骗等信息严重困扰人们的生活。针对朴素贝叶斯对信息分类时考虑所有特征并将特征赋予相同权值两方面的缺陷,提出一种用于微信信息分类的改进贝叶斯算法。采用改进的互信息进行特征选择,提取关键特征,通过改进TFIDF对特征加权,优化朴素贝叶斯的分类性能。实验结果表明,改进的贝叶斯算法能有效选择关键特征属性,提高微信信息分类的精准度。

The prevalence and development of WeChat has facilitated the dissemination of information.However,it also comes with advertising and fraudulent information,which brings a lot of troubles to people＇s lives.For the two important factors that naive bayes considering all the features and characteristics are given the same weight.We put forward an improved bayes algorithm.Selecting features by improved mutual information and weighting by improved TFIDF to optimize the bayesian classification performance.Experimental results show that the improved bayes algorithm can effectively select key features and improved classification precision