Research And Application Of Mobile Application Recommendation System Based On Time Context

In recent years,the Internet,especially the mobile Internet,has grown rapidly and updated technologies frequently.Smart mobile devices,such as smart phones and panel PCs,have become more and more popular and led to a dramatic increase of users who possess them.As an important part of smart mobile devices,mobile applications have changed the way people live,work and learn,and have occupied a significant place in the design and development of system.In some mobile app markets,such as Google Play and Apple Store,the amount of mobile apps has reached one million.Such huge mobile apps do not only provide convenience for users and references for developers,but also bring new challenges.From users' perspective,it is very difficult to find applications that they are interested in.From developers' perspective,selecting the appropriate applications as references will cost their vast energy.At the same time,the trend of lowering ages of smart phone users cannot be ignored.And how to keep younger users from indulging in the smart phone is also a problem to be solved imperatively.The thesis researches about personalized recommendation for mobile applications.First,improved the recommendation algorithm,proposed a resource diffusion algorithm based on user division.After researching personalized recommendation algorithms,it can be found that preference transition accumulated from time dimension had a great effect on the accuracy of recommendation algorithms.To improve traditional resource diffusion algorithms,the thesis,on the basis of graph-model resource diffusion algorithms,proposed a new resource diffusion algorithm based on user division by introducing time context information to the algorithms.The improved algorithm regards users of preference transition from time dimension as multiple users,namely,under the thought of user division,it introduces time context information to utilize data information more fully.By comparing experiments,it was found that the improved algorithm had much higher accuracy than the traditional algorithms.Next,the thesis applied the resource diffusion algorithm based on user division to the recommendation of mobile applications,designed a mobile application recommendation system for smart phone users and developers respectively,and realized the system prototype.When recommending mobile applications for users by personalized recommendation algorithm,the system not only considers the users' preferences but also refers demographic characteristics from the user's personal information such as age and gender to provide different recommendations strategies for users.In order to find reference information in a convenient way for developers from massive mobile apps,it provides users' portrait for application,the function of application query based on user group feature and correlation between mobile applications.Finally,considering production,growth and decline of mobile applications,maintain system in the view of the information theory.