A new method to identify blog comments spam was proposed. The short comments were identified by the network common words first, and made K rounds to identify the comments which used the improved similarity formula. Following every identifies, the weights of keywords and extend keywords were adjusted. All the comments were identified to the category. The spam reviews were filter again by the network common words and the keywords, and more legitimate comments were identified. Experimental results showed that the method, to some extent, improved the recognition accuracy.%针对博客垃圾评论泛滥的问题,提出了一种识别博客垃圾评论的新方法.利用网络常用语对短小评论先进行评论的识别,然后利用改进的相似度公式对评论进行了K轮评论的识别,在每轮识别之后,对主题词进行权重的调整和主题词扩展;待所有评论识别完毕,再利用网络常用语和主题词对识别出的垃圾评论进行第二次过滤,过滤出垃圾评论中的合法评论.实验结果表明,利用该方法进行评论识别在一定程度上提高了识别垃圾评论的准确率和召回率.
展开▼
