Abstract: According to the feature of video-sharing with P2P in campus network, a video recommendation system in campus network called FP-CNVR system was designed and implemented; a novel data cleaning method based on customer segmentation was proposed; the structure of FP-tree in the process of frequent pattern mining with FP-growth algorithm was improved. Experimental results show that the novel recommendation system can provide more reliable recommended results, and also the novel data cleaning method can improve the accuracy of the recommendation results. The research may have some inspiration to related subjects in campus network.