A Study On Context-aware Collaborative Filtering Recommendation Model In Mobile Environment

With the rapid development of mobile Internet, the mobile client ’s products and services areincreasingly abundant. The smartphone as a representative of mobile devices have become a part ofpeople’s daily life, work and entertainment. However, the limited size of the mobile terminalequipment, the limited capacity of mobile users to receive information and the explosive growth ofinformation in mobile environment makes it difficult for mobile users to look for an effectiveinformation. Just like traditional Internet, the information overload problem become increasinglyprominent.Recommended system is a good choice and has had outstanding performance in solving thetraditional Internet overload problem and helping users to make decision according to theirpreference, especially collaborative filtering algorithm has become the most successful and widelyused algorithm. Therefore, we consider that we should introduce the collaborative filteringalgorithm into the mobile environment. However, the new features of the mobile environmentdetermine the traditional recommendation mode can not be totally used in the mobile environment.Context has become an important role in making decision for mobile users.This paper presented a model of context-aware collaborative filtering recommendation inmobile environment, and introduced context factors to collaborative filtering recommendationalgorithm to provide users with a better recommendation quality in a mobile environment. Firstly,we elaborated the importance of context in mobile recommender, and proposed sparsity problemfaced when the context was introduced into collaborative filtering recommendation and the shortageof context introducing method. For solving sparsity problem, this paper defined the relaxed contextto improve the data sparseness problem that was caused by the introduction of context. Thenaccording to the shortage of context introducing method, we proposed a method that combinedcontextual pre-filtering with contextual modeling. Corresponding to the contextual pre-filtering,we defined a rigid context as a pre-filter limitation. Corresponding to context modeling, this papercame up with an idea of an improved context-based collaborative filtering recommendationalgorithm based on traditional collaborative filtering recommendation algorithm. Then weproposed the idea of context-aware collaborative filtering recommendation model and framework,and made a detail description of the core algorithm of the model. In addition, this paper used asurvey to build an experimental data set, and designed several experiments to compare the qualityof the recommendations proposed algorithm with the other three algorithms with MAE and F1.Experiment result showed that the proposed model had a better performance in the recommendation quality compared to the others. Finally, we summarized the research work that was done by thispaper, and summarized the insufficient of this model.At last, the future direction for furtherresearch was also listed.