Study On Mining Technology For Web Viral Marketing

With the development of E-commerce, Web viral marketing is being concerned and applied more widely according to its unique advantages compared with traditional ways. At present, data mining has been successfully employed in traditional marketing; however, there is little work on viral marketing.First of all, a detailed analysis is done on advantages and disadvantages of the models or algorithms that have been proposed. In order to meet the demands of data mining for viral marketing, this thesis defines Customer Trust Network (CTN), and builds a model for Diffusion of Product Information based on CTN (called CTNBDPI). The model takes into account simultaneously internal and external factors that influence on customers'adoption of products or services, and solves the issue on isolated nodes'acceptance and recommendation. Meanwhile, two corresponding algorithms for mining the core group—VMCGM (Viral Marketing Core Group Mining) and E-VMCGM (Enhanced VMCGM) are raised, together with their complexity analysis. In addition, this thesis investigates the global optimization on viral marketing and suggests a kind of solution.Furthermore, dynamic is one of the essences of social networks, which is omitted by present work on mining technology for viral marketing, making dynamic viral marketing strategies unavailable. In order to solve this problem, this thesis proposes an algorithm based on Network Change Log (NCL) to discover core group, which is called DMCG. The algorithm can avoid re-computations on the changed CTN effectively and guarantees the accuracy of mining. Besides, it proposes new ideas on how to control diffusion of negative influences in a network.Thirdly, the proposed models and algorithms are inspected with experiments. The results show that the model can be in line with the spread law of information in viral marketing, and the algorithms have better time and space complexity. Compared with previous models, the proposed model in this thesis is helpful of making more effective viral marketing strategies.