Research Of One Class Collaborative Filtering Algorithm Based On Real-time Information For E-commerce

While the development of the Internet offers users convenience and surprise, it also brought them confusion and frustration. On one hand, the users enjoy a wealth of information resources; on the other hand, it is difficult to quickly find the information they need. With the rise and development of e-commerce in the late20th century, personalized recommendation technology developed rapidly and is applied to e-commerce recommendation system quickly. And it plays a role which can let users avoid getting lost in the mass of information, and it can help users make decisions, select the desired items, improve sales, In recent years, with the advent of "big data era", information overload and information explosion become frequent, collaborative filtering of personalized recommendation technology is becoming more mature and successful. However, in many cases, the data what e-commerce systems collected usually states user’s preference clearly, without considering the data which not state user preferences clearly or not state at all, the result lead to the recommended range of small, and not accuracy. The method which uses the latter data to recommend is One Class Collaborative Filtering referred OCCF.Compared with collaborative filtering, research on OCCF is less. Sparse data give the study much difficulty. But it also shows that OCCF has potential research value and space. Based on the study of a large number of relevant literatures, the article proposes a solution for the data sparsely and real-time problem. First summarizes the basic research method in this area, then uses the singular value decomposition, the weighting matrix approximation methods, designs an OCCF recommendation algorithm, and establishes an appropriate model. In weight setting, based on previous studies, introduce real-time information:users recent browse information related to CRM and the goods to market information which related to PLC. Finally, makes simulation of the design of algorithm with Matlab, and compares the algorithm with other basic methods, verifying its effectiveness.Proven, the OCCF recommendation algorithm based on real-time information is superior to the other. Currently, the product life cycle gets shorter; user buying preferences changes over time, using the recommended method can effectively improve the recommendation quality of e-commerce recommendation system. It has positive meaning for e-commerce in cross-selling, personalized, targeted recommend market.