Research On The Comprehensive Recommendation Algorithm Based On The Dynamic And Static Properties Of The User

With the development of the Internet, people are gradually entering an era of information overload, while accompanied by the emergence of two major problems for the user how to find the information they are really interested in, and how to find more information to make it of interest to the user. The emergence of the recommendation system is in order to solve these two problems. Collaborative filtering algorithm is the most successful recommender system and the most mature algorithm, but there are sparse data accuracy issues and collaborative filtering algorithm with the recommended quality greatly reduced, while the cold start can not be ignored. To solve these problems, the paper proposes a combination of online and offline nearest neighbors of user behavior data comprehensive search based on user research dynamic and static properties of the composite recommendation algorithm.In this paper, the main research results are as follows:(1) related technologies in-depth study of recommender systems, and analyzes the advantages and disadvantages between the commonly used recommendation algorithm, through its existence leads to the deficiency of the user based on the users the importance of static properties of comprehensive recommendation algorithm.(2) in a new way to classify the user attributes, the normal user attributes are classified into static and dynamic attributes, and the method of calculating the similarity between users is modified by dynamic and static attributes. The advantage of this method is that when a new user can enter the system, according to its static properties to predict the properties of the user's preferences, to a certain extent, ease the "cold start" problem.(3) in order to get rid of the shackles of conventional online data collection, this paper presents a data collection algorithm to predict the properties of user preferences, the user line transformation to collect information and user preferences for the attributes for complete bipartite graph, through complete bipartite graph min max to set algorithm, to achieve user preferences for attributes of mining, and its classification for static properties. (4) the use of dynamic and static properties of dynamic updating model to simulate the dynamic and static properties of real-time transformation processing, more accurate positioning to the taste properties of the target user, effectively solve the "sparse data" problem to some extent, inally, the experimental results show that the algorithm. The results show that:based on user recommendation algorithm for integrated static property can be more efficient for the user recommendations, not only to find the nearest neighbors to improve accuracy, but also to solve the "cold start" and the great "sparse data" problem improve.