Context Awareness Of Recommender Systems Research

With the explosive growth of Internet information resources,it is increasingly difficult for users to find out resources meets the needs,and this formed the problem of the information overload.Information overload is the side-effect of too much information resources in an era of "big data".The recommendation system is regarded as a kind of effective method to relieve the information overload problem.The recommendation system pushes the most likely product or service to meet the users' needs,by collecting their information and analyzing their behaviors,and using personalized interest modeling or machine learning methods.Since the recommendation system has advantages of initiative service,personalized push,strong practicability,the system has widely applied in practical applications such as e-commerce,digital library,and other fields,and has become the key research direction in the field of application.As the user demand for personalized experience is increasing in recent years,and multi-platform application development allows the recommendation system to expand unceasingly,the application scope of the current recommendation system is more and more difficult to meet the needs of users.In addition,the rapid growth of the information also leads to produce a large amount of complicated data in information research area.The traditional way of data research cannot have quick and effective access to the essence of law behind the data when confronted with the huge amounts of data,the process of research consumes a lot of time and human resources.In the light of the above problems,this paper puts forward a recommendation system which is based on the technology of context awareness.The system is based on the traditional recommendation technology,combined the technology of scene perception,while fully considering many platform features to improve the accuracy of the recommendation algorithm.At the same time,this paper uses data visualization technology to show recommend list in the form of the chart,makes recommendation results more intuitive and easy to understand,improves the user experience of the recommendation system.The main research content of this article is as follows:Firstly,this paper summarizes the background and research current status of the recommendation system,and then mainly reviews the content,the principle,the process,the advantages and the disadvantages of recommendation which based on the content and the collaborative filtering recommendation algorithm.At last,this paper analyzes the main issues of the current recommendation algorithm.Secondly,this paper explains the concept of context awareness,and puts forward the concrete application of context awareness technology in the recommendation system.Then,it puts forward a context awareness of collaborative filtering recommendation algorithm.The algorithm which based on the "user-product-scene " three-dimensional tensor,imports the information of context awareness in a creative way while fully considering the platform features and the accuracy,it also transforms the traditional collaborative filtering algorithm into the context awareness of cooperative concern algorithm,which effectively reducing the grading error of the prediction.Finally,by the contrast experiment,this paper verifies the feasibility of the context awareness recommendation algorithm.Thirdly,this paper changes the research method of the recommendation system,from the traditional "user-product" mapping to rate,to the new"user-goods-scene" mapping to rate.It also makes it more diverse to research the content of recommendation system.At the same time,this paper solves the problem of prediction score of the recommendation system by using tensor decomposition.Finally,this papers imports the data visualization technology to the research of recommendation system,also summarizes the background and the general steps of the data visualization technology.Then,it programs the context awareness of recommendation and data visualization system.The system has two major functions,one is prediction score with fixed user?product and context,the other is generation recommended list with fixed user and context.Finally,the system shows the result of the experiment data in the form of a visual chart.