A cross collaborative filtering recommendation method

Abstract

The invention relates to a movie recommendation method, in particular to a cross collaborative filtering recommendation method. According to a project-based collaborative filtering algorithm (Intem-based CF) principle, an inter-movie similarity calculation strategy is designed, and a serialized similarity dictionary of each movie is solved; model-based collaborative filtering algorithm (Model-based CF)- A Slope One algorithm is combined with a local similarity principle to carry out scoring data filling on an unscored movie, so that the sparsity of a scoring matrix is greatly reduced; and on the basis, providing a user comprehensive similarity calculation model fusing the score similarity and the average weighted interest similarity among the users, and then finishing movie recommendationof the target user by applying a user-based collaborative filtering algorithm (CF). Compared with similar methods, the movie recommendation accuracy can be remarkably improved.

Description

technical field [0001] The invention relates to a movie recommendation method, in particular to a cross collaborative filtering recommendation method. Background technique [0002] People live in an era of big data in which information technology is changing with each passing day, mobile Internet and cloud computing are developing rapidly, and the amount of global data is exploding. Compared with the traditional text-based structured data that is easy to store, the proportion of unstructured data such as audio, video, and pictures has gradually increased. The rapid growth of information on the Internet, on the one hand, enables people to obtain more and more information resources, which brings great convenience to people; on the other hand, facing the massive information resources, people have to spend more time and energy To search for helpful information, the problems of information overload and resource obsession are becoming more and more serious. [0003] Recommendati...

AI Technical Summary

Problems solved by technology

However, from the research status of collaborative filtering recommendation algorithm, we can see that the traditional collaborative filtering recommendation algorithm has the following problems: data sparsity, users voluntarily give few evaluations when there are any incentives, which makes the calculation of users or items difficult. The similarity between them is not accurate enough; the cold start problem, when a new user or project enters the system, the algorithm cannot handle it well; the scalability problem, the algorithm does not handle big data well, and the real-time recommendation is very poor Difficult to guarantee (see "Research Progress in Collaborative Filtering Recommendation Algorithms" published by Weng Xiaolan and Wang Zhijian, [J]. Computer Engineering and Applications, 2018, 54(1), P25-31)

Images