The Design And Implementation Of Personalized Recommendation System Based On Collaborative Filtering

With the rapid development of mobile Internet and the popularity of mobile devices,more and more people cannot live without mobile phone,tablet and other mobile terminal equipment.As for the greater demand for people through mobile devices quickly and accurately obtain the useful information,therefore recommend technology applied to the field of mobile is also expected.Now the mobile phone and other mobile devices is the main way for people to obtain information,compared with the traditional Internet devices.People's hobbies,consumption habits and more personal information are more easier to be explored through mobile phone.Therefore,the personalized recommendation based on mobile terminals is a hot topic at present.What is the most successful technology in the field of the recommendation is the collaborative filtering technology.Nowadays,the recommendation technology in the commercial domain is not only applied to the collaborative filtering technology,but also it often combines the collaborative filtering with other recommendation technologies,which can form the more accuracy of the hybrid recommendation technology.This paper proposes an improved recommendation algorithm to realize the mobile user's personalized recommendation system based on the traditional collaborative filtering.The main innovations are summarized as follows:1.A personalized recommendation algorithm based on forgetting curve and user access weights is proposed.The algorithm has the following improvements: firstly,the initial user-scoring matrix is processed by the forgetting function;Secondly,the neighbor set is determined by the improved pearson coefficient to calculate the similarity between users;finnaly,recommend to the target user by the rating prediction algorithm.2.Establish a user model and message model for O2 O application in community.User model mainly includes the user's interests,preferences and related operation behavior.This paper includes the user's interests and operation behavior in the client of the news posts;the message model mainly includes the classification,the accesses,user's reading time,and whether the user can transfer the information or not and so on.Finally,using the information to provide the corresponding data support for personalized recommendation.3.Adopting the framework of Struts+Spring+Hibernate,and using this technology to design and implement the server,and also add a personalized recommendation module in the server so that we can easily get the user's relevant interest preference datas and related behavior datas from the database to make the corresponding personalized recommendation.4.In mobile client,we use the MVC design pattern to complete the design and implementation of the mobile client so that it can display the contents of personalized recommendation.The layered design makes the structure of the application more clear,the coupling degree more lower and the expansion more convenient.