Recognition of Spam Microblog for TV Program Evaluation Under Mircoblog Platform

Aimed at the serious problem of the microblog platform used in the field of evaluating TV program that is badly affected by spam microblog, this paper proposes a recognition method about combination of lexicon match with SVM based on pattern matching and machine learning. At the same time, considering the impact that spam information caused in the public-opinion-trend and topic-attention-degree, it is important to identify the spam microblog correctly. They are various cleaning modes for different spam information. And the results of experiment shows that the total-recognition-rate has already reached 80 %. This method is useful for the following text mining.