网络博客空间中基于半监督学习的垃圾评论检测

针对网络博客空间中垃圾评论泛滥的问题，给出一种半监督学习式网络垃圾评论检测方案。基于评论内容的统计分析，设计相关度、词组重复率、超链接数目、内容淫秽度、句子长度共5个特征指标，给出网络垃圾评论检测系统的框架，并进行实验验证。实验结果表明，本方法能有效检测出网络博客空间中的垃圾评论，具有较好的应用价值。

Detecting Techniques Based on Semi-supervised Learning for Comment Spam in Internet Blogsphere

To identify comment spam which has flooded in the blogsphere,a detecting solution for comment spam was proposed by using immune optimization.The key features of comment including content-related validity,phrase recurrence rate,number of hyperlink,obscenity,and length of sentence was expound,a frame of immune optimization algorithm was given,the framework of detecting system for comment spam were designed,and simulation experiments were done to validate our detecting solution.Experimental result shows that detecting solution given in this paper can identify comment spam availably,and has the advantage of good application value.