Research Of Commodity Recommendation Model In B2C E-commerce

With the development of information and internet technology, E-commerce becomes more and more popular as an efficient new commercial model. E-commerce brings people more choices, but it also brings commodities overload. This situation makes it difficult for consumers to find the commodities which they need rapidly. Because E-commerce website provides enormous products and separate customers from real commodities. In this background, Commodity Recommendation systems emerge, which can recommend satisfying commodities to customers by customer's character and the correlation between commodities.This thesis particularly chooses B2C E-commerce website as a respective study object for commodity recommendation model. Base on analyzing the characteristics of different recommendation technologies, design a commodity recommendation model founded on frequency matrix and text similarity, which suits for B2C E-commerce website. This thesis studies data analysis and extraction from Web log, commodity correlation finding based on frequency matrix and text similarity computing. Firstly, introduce the connection between personalized service and commodity recommendation; compare different recommend technologies and their adaptability for B2C E-commerce website. Secondly, design a commodity correlation finding algorithm founded on frequency matrix and use the commodity correlation to recommend commodities. Thirdly, use text similarity computing method to compute commodities' similarity for improving this recommendation model's cover and matching rate. At last, show the recommendation system's technical scheme, system structure and specific method.Base on the theory, a Commodity Recommendation Prototype System is developed, which contains six modules: Data preprocessing module, commodity correlation finding module, commodity recommended simulation module etc. Experiment shows that the model realizes commodity recommendation service's functions, it has preferably real-time performance and practicality, achieves the expected results.