Pre-Filtering Contextual Recommender Technologies Based On Double-Splitting

Nowadays, we live in an age of information explosion. The rapid development of information technologies have great impact on social life, and the amount of information we can obtain becomes more and more. It brings great convenience to people’s life but also brings the problem of information overload. People found it need to spend a lot of time to find the real information from internet. Therefore, how to help users find the real interest information in the shortest time is very important.Recommendation system can be used to solve this problem, and the user’s personalized information can be got through the data set. According to the user’s preferences and needs, it can help users find their real interest.In recent years, the related research on recommendation systems has been obtained widely attention of scholars at home and abroad, and the algorithm of recommendation as its corn content has become maturer and mature. People began to try to further improve the recommendation from the correlation effect. For example,adding the context information to the process of generating recommendation and Context-aware Recommendation technology has become one of hot research branch of the recommendation system. Therefore, it is important to take the context of factors into the research and applications in the process of recommendation. This paper makes a large amount of research work based on the direction. In view of the present defects of known algorithm, we put forward two recommendation algorithm based on the idea of double-splitting and combined with the real data sets on the validity of the algorithms is verified. The main contents of this paper are as follows:(1) According to the existing context-aware recommendation we propose a context-aware recommendation algorithm based on the Double-splitting. In order to get a more conducive of the prediction data subset scores through the algorithm, we give a second time on the decomposition of the data set in a fixed threshold range.We compare the algorithm and other classical recommendation algorithm based on the public data set and verify the effectiveness of the new algorithm.(2) This paper proposes a context-aware recommendation algorithm based on Hybrid-splitting combined with the barter transaction context information. The algorithm considers the integration of context information of user and objects, and itreduces the losses caused by the combination of decomposition through introducing error correction function. In the end, we conduct a series of experiments on barter transaction real data set to validate the new algorithm.