A clustering method based on concept was provided to analyse the sentiment polarity for Chinese Bloggers. The concept is introduced into Vector Space Model (VSM) on the basis of HowNet. Firstly, sentiment words are extracted from blog texts which would be expressed by VSM with the concept of sentiment words. Secondly, blog texts are clustered with k-means algorithm to finish the analysis of sentiment polarity for Chinese Blogs. The precision of sentiment polarity analysis of Chinese Blogs is improved with concept as feature in VSM. The experiment proves the concept based VSM to be of better performance than traditional term based VSM in clustering analysis of Chinese Blogs on sentiment polarity.