基于朴素贝叶斯的伪装型垃圾网页检测  

Cloaking detection based on Native Bayes simple

作　　者：王莉丽[1] 朱焱[1] 马永强[1] 

机构地区：[1]西南交通大学信息科学与技术学院,成都610031

出　　处：《计算机应用》2013年第A01期102-103,106,共3页journal of Computer Applications

摘　　要：针对伪装型垃圾网页的隐藏性、欺诈性和难以检测等问题,运用二元分类方法对伪装型垃圾网页进行检测。该方法将网页分解成内容特征和链接特征,并采用朴素贝叶斯(NB)算法对伪装型垃圾网页进行分类检测。最后,将NB算法和几种常见的分类算法在同一数据集上进行测试,结果表明,朴素贝叶斯算法的综合评价F1值达到94.64%,明显高于其余几种分类算法,能够有效地检测出伪装型垃圾网页,提高网络信息检索质量。Aiming at the problem of hiding, deception and difficulty to detect cloaking, the paper applied two-class classification method. Webpage was decomposed into content characteristics and link characteristics, and Naive Bayes （NB） algorithm was used to detect cloaking. Finally, this paper used the NB and several common classification algorithms to classify the same dataset. The results show that the F1 value of the NB algorithm is up to 94.64%, which is obviously higher than that of other kinds of classification algorithms. The method can detect cloaking effectively and improve the quality of network information retrieval.

关 键 词：伪装型垃圾网页检测 机器学习 分类算法 朴素贝叶斯 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术] TP393.098[自动化与计算机技术—计算机科学与技术]