Research And Implementation Of Accurate Information Recommendation Technology Based On User’s Interest Model

Users face the grim problem of information overload when it becomes the era of the big data. The technology of Information recommendation can help users extract useful information from the huge information on the Internet, thus attracting a wide range of researchers’ attention. Unfortunately, most of the existing recommendation technologies recommend hot information, which satisfies users’ individual information needs, to users by considering the commonality and people’s interest. In other words, they recommend by considering from the perspective of universal connection and they consider less the specific characters of interest of people. That is to say, they consider less from the perspective of individual demands of users. However, recommending information through considering people’s personalities is extremely important, especially in the process of recommending information where the information of users’ preferences is needed, such as recommending online mall book, movies in Netflix and network news.Therefore, in order to solve the problem of information overload and recommendation through users’ personalities, this paper studied the user-based and the item-based collaborative filtering algorithm. This paper discovered the advantages of implicit model through analyzing and comparing the experimental performances of implicit and explicit models, and realized the personalized recommendation system by using the collaborative filtering algorithm based on the implicit model. The main research work of this paper is displayed as follows:Firstly, this paper studied the algorithm which uses explicit characteristics to express interest of users. Explicit characteristics are the information of categories and careers of people and items in the data set. Based on explicit information, this paper fitted the users’ interest and rating data using the linear regression algorithm, then predicted ones rating score to a movie by utilizing the weighted parameters, and then realized the information recommendation though the rating scores.Secondly, this paper researched the fusion model to represent the users’ interest and hobbies, and realized this recommendation algorithm that is based on the interest fusion model. The interest fusion model is the integration model of explicit and implicit characteristics where the implicit factors are the features that are trained by latent factor model. The recommendation algorithm based on interest fusion model took both the implicit and explicit interest models into consideration, and can simulate preferences of different users’ personalized features via adjusting the weights in the experiment.Once again, this paper studied and implemented collaborative filtering algorithm employing implicit characteristics. It expressed users’ interest by using the implicit models, and then this paper implemented the collaborative filtering algorithm based on users’ implicit characteristics. Implicit characteristic is able to describe the interest of a user more exquisitely and builds the user’s interest model more precisely. As a result, it makes the recommendation system can grasp the users’ personalized interest demands more accurately, so it can realize personalized recommendation for users.Finally, the paper realizes personalized recommendation system utilizing the above algorithms, and testes the performances on the data sets of MovieLens, the experimental results show that the implicit characteristic can state the users’ interest more precisely than explicit characteristics do.