The Research And Application Of Collaborative Filtering Algorithm In The LBS Community Service System

With the deeply implementation of reform and opening policy, and the development of technology, the development of community comes into a new times, The introduction of technology is an effective way to improve the quality of community service. Most of the current community service sites stagnate in the Web2.0era, and currently the most widely used big data technology and mobile Internet technology still not used to the community service.The aim of this project is to design and develop a LBS community service system which based on the mobile terminal. The system aimed at solving currently community service problems such as the lack of interact between residents, the low level of personalized service. For the scenarios of LBS community service, we try find a method to optimized recommendation result, and also provide a way to solve "cold start" problem through location information which is the unique characteristic of mobile internet. The thesis proposed a method, based on the user’s location, to divide user. This method divides user into group by searching TDT(two-dimensional tree). According to the result, we get statistical information of every group, then add the information to ordinary matrix. Recommendation engine, which using ALS-λ. algorithm, provides the recommendation result based on adjusted matrix.The thesis test improved algorithm by actual data. According to the result, we show the feasibility and effectiveness of the improved algorithm. In actual application, the improved algorithm can also to some extend the "cold start" problem of recommendation system. Recommendation engine can also provide results of the novelty when user changes their location. The main work of this thesis is described as follows:1. Analyze the requirement and difficulty of community service system, we choose mobile app as application form. Design the LBS community service system which based on the mobile terminal according to need.2. Deeply research collaborative filter algorithm. By collecting and researching related algorithm information, we propose and implement an user clustering method which based on location information.3. Research some open source recommendation project, we improve the recommendation algorithm with user clustering information. Extract the data needed from user behavior data, and verify the effectiveness of the algorithm based on actual data.4. Implement recommendation engine, and show the recommenda-tion result.