The Research Of Recommender System Based On Context-Awrar

With the popularization of Internet,the resourses of network is more and more rich,at the same time,the data is growing exploding.The problem how to get the information is not what people confused,but how to dig out the useful information which we need from the vast amounts data is the urgently problem.Personalized recommend system arises at this moment,it need to present different outcome list based on the needs of the different suers,as a result,the context information has been widely used in different areas,and the context information with the recommend system can enhance the accuracy of recommendation,to help users find the outcomes that are fit with the needs,however,now the research of recommended system is still at the primary stage,there are still many problems need to slove in the future.In this paper,we aiming at the research of the recommended system which has existed and the context information,and put the context information into the recommended system.In this paper,the main contents including:the history of the recommended system and the research results,the principle and the method of the traditional recommended algorithm,and the successful application of recommended algorithm in different areas,the history and the research results of the context awareness,the basic concept and the filtering of the context awareness,and the process of data.As a result,we propose a improved recommended algorithm based on the traditional recommended algorithm to enhance the accuracy of recommendation.This paper also applied the improved recommended algorithm to the video recommended applications,and validate the effect of the recommended algorithm.The research results of this paper is as followed: (1) The improved recommended algorithm.In this paper,based on the model of SVD and the model of user-user,we combine the two models,and add the factor of time,and add the context information,such as gender,age and occupation,to optimizate the recommended algorithm,and using the dataset of MovieLens to verify the validity of the recommended algorithm, the dataset of MovieLens includes training and testing set,finilly,we compare the various recommended algorithm and the improved recommended algorithm,the result shows the improved recommended algorithm has a good effect and the RMSE has steadily declined.(2) The video recommended application.This paper designed a video recommended application,which applying the improved recommended algorithm into the video recommended application,mainly including the recent update module,the popular video module,the video recommended module(including TV channels,movie channels,sporys channels,the show channels and cartoon channels),research module and comments module,which means the success of the improved recommended algorithm.