With the rapid development of mobile communication technology and mobile Internet,intelligent mobile terminal equipment became popular.Various types of mobile applications were also exploded.Users could not locate and acquire favorite applications quickly and accurately in mass applications.A large number of excellent applications were also covered under popular mobile applications.In order to help users find their interested mobile applications efficiently,and provide more accurate application services.At the same time,long tail application could get more exposure opportunities.Personalized application of mobile applications came into being.The existing mobile application recommendation algorithm relied mainly on the user's behavior of using mobile applications.They did not fully consider the user's complex context.Simultaneously,the introduction of context information made the sparseness of the recommendation system more serious.In view of the above problems,this paper presents mobile application recommendation algorithm based on user context similarity(UCA-TF algorithm).It integrates the complete context representation into the recommended model and rebuild the user context similarity model.What's more,the tensor decomposition is used to process high-dimensional sparse context data.It's a great help to find interesting mobile applications quickly,and improve recommended accuracy and mitigate data sparseness effectively.Finally,based on the mobile application recommendation system reference model,a design and implementation scheme of mobile application recommendation system based on UCA-TF algorithm is proposed.The main work of this paper includes:(1)This paper makes a literature review of mobile recommendation system and context-aware recommendation.It summarizes the characteristics of mobile application recommendation algorithm,the key technology and the problems to be solved urgently.Finally,it describes the research background and significance of mobile application recommendation system.(2)It aimes at the complexity and sparseness of user context in mobile service recommendation,mobile application recommendation algorithm(UCA-TF algorithm)based on mobile user context similarity is proposed.Multi-dimensional context similarity model is established with combining the user context similarity andconfidence of similarity.Then,K-neighbor of the target user information is applied to the three-dimensional tensor decomposition,composed by User,Context and M-application.Therefore,the predicted value of the target user is obtained.The experimental results show that the algorithm proposed in this paper has higher recommendation accuracy and can effectively mitigate the influence of score sparse.It also improves satisfaction of user.(3)Based on the study of UCA-TF algorithm,the design and implementation scheme of mobile application recommendation prototype system based on user context similarity is proposed.Mobile application recommended client displays application recommendation information.The user's request is responded in real time.And client uploads the user behavior data and the context information collected to the server.The server obtains the stored user history information,carries on offline data calculations and generates a pre-recommended list for call.In the meantime,the server manages and maintains user information and mobile application information.So the mobile client and server-side content can be updated and adjusted in a timely manner. |