Binary Attribute Clustering-based Personalized System

Web Data Mining technology is introduced in this paper at first and then the clustering analysis which belongs to Web Data Mining area is discussed in detail which focuses on the classification attributes of data clustering. In view of Web log clustering, a two-value property is used to describe the time in Web log and then Zipf s law is used when the time is converted to two-value property. ROCK clustering algorithm is introduced in this paper and Jaccard coefficient is always used as similarity when we use ROCK algorithm. The characteristic of similarity value calculated in this algorithm is that the bigger the similarity value between the two objects is, the more similar the two objects will be, that the smaller the similarity value between the two objects is, the more different the two objects will be. Because the calculating cost of ROCK algorithm is relative too high, the algorithm is mended in this paper. According to the characteristic of similarity, the similarity value is sorted in descending order, and then these objects are clustered orderly. The improved ROCK algorithm can get the clustering result quickly and precisely.In view of the individuality service demand in Web-based learning, the individuality Web-based learning model is constructed as well as the learner individuality information model. A personality recommendation system of e-learning (PRELS) is designed and implemented in this paper with thought of Web-based learning individuality recommendation. This prototype system mainly applies the Web Data Mining technology to the environment of Web-based learning。Combining log mining and clustering analysis, a personalized recommendation algorithm is conducted.The system makes recommendation for learner with study resources, study behavior or the study shortcut, improves network study quality, enhances the achievements of users for studying on network study, and also meets the learner's need in individuality study.