Research and Application of Web Recommendation System Based on Cluster Mode

Web recommendation system is an important research content of web mining. In this paper, we propose a new web recommendation system model based on cluster mode to realize the real-time online recommendation. First, we use a new method to get the feature vector based on tf-idf method. Second, we use an unsupervised web page clustering algorithm to realize user clustering. According to the result of clustering, we use naïve Bayesian method to predict user´s action according to its web navigation. Experimental evidence shows that using this method to explain users´ active browsing goals is effectively enhanced.