The Design And Realization Of Active Pushing System Of Personalized Consumption Information Based On Data Mining

With the development of IOT, such concepts as intelligent home, intelligent transportation and intelligent medical etc, have been proposed recently. And today people’s life is getting more and more intelligent. In IOT, one of the most important technologies is sensing technology. It collects information of items by using RFID, infrared sensors and other sensing equipments and transmits to the platform. How to effectively analyze and process the data information collected has always been the focus of researchers. To solve this problem, this thesis designs a system that applies the idea of data mining to the process of data information, and pushes the appropriate recommendation result to users actively.This thesis first introduces some existing technologies in intelligent health service, and some background knowledge of data mining. After analyzing the characteristics of the data collected from the wireless sensor network, this thesis preprocesses the data, such that the input data meets the requirement of data mining. This thesis designs the recommendation algorithm based on the collaborative filtering. Using the algorithm designed, we get the recommendation information for users, i.e. “What to Push” in the pushing system. In the second part, this thesis introduces some pushing technologies and compares the differences between traditional “Pull” technology and “Push” technology. The thesis then analyzes the pushing demand of the system and determines that it realizes the function based on AJAX long polling, i.e. “How to Push” to users.In the end, this thesis realizes the active pushing system of user service by Java and database design. The whole system adopts the B/S three-layer architecture and deploys on the server. Users interact with the system through the browser interface. When the wireless sensor network collects and transmits the health data information of the user, the server processes the data by collaborative filtering algorithm and searches for the neighborhood users of the user, and then recommends health service they chose for the target user. Finally, the server pushes the recommendation result to the browser actively.