Abstract：A public sentiment monitoring system is designed, based on co-ICIB co-clustering.The system collects the information from micro blogs and blogs, creates public sentiment database and analyzes the hot points and development trend of public sentiment by such data mining algorithms as co-clustering.Once hot points are confirmed, which means they become truly concentrated topics in Internet, text categorization algorithms could group the information of the same topic into one category and help the track trend of public sentiment.The results of the system could provide the original public sentiment data, data chart and suggestion analysis for a public sentiment supervision department.

Keyword：We Media;public sentiment;co-clustering;