Rating prediction and recommendation method of recommendation system based on user behavior tendency probability

Abstract

The invention provides a recommendation system scoring prediction and recommendation method based on user behavior tendency probability for scoring and calculating the relationship between an item, a label of the item and the user's tendency scoring of a current recommendation system, and providing a similar item recommendation service for a target user when the target user searches for items in the current recommendation system. The method is characterized by comprising the following steps: 1, building an item database; 2, building a user rating library; 3, calculating the tendency probability prediction scoring P(ur, Tq) of a user ur towards a label Tq; 4, calculating the tendency scoring P(ur, Ii) of the user ur towards an item Ii; 5, calculating the similarity S(Ii, Ij) of the item Ii and an item Ij when a target user u* searches for the item Ij from a current recommendation system; 6, calculating a prediction scoring value r(ur, Ii) of the target user u* towards the item Ii; 7, sorting neighbor items according to the prediction scoring value r(ur, Ii), and recommending the neighbor items ranked in predetermined numbers to the target user u*.

Description

technical field [0001] The invention relates to a recommendation method, in particular to a scoring prediction and recommendation method of a recommendation system based on user behavior tendency probability. Background technique [0002] As a content of personalized service, the recommendation system can recommend items such as movies and commodities that users are interested in or suitable for themselves based on the similarity between items, so that users can quickly obtain information about items that match their preferences and preferences. At the same time, the recommendation system can save the cost of system resources and bandwidth consumed by users in the process of browsing a large number of items, making the recommendation system trusted, favored and used by a large number of users. [0003] In the prior art, the recommendation system mainly adopts the collaborative filtering recommendation method, which utilizes the common preferences of user groups with similar ...

AI Technical Summary

Problems solved by technology

On the one hand, the traditional recommendation method only relies on the similarity between users to select the neighbors of the target user, resulting in low recommendation accuracy and poor diversity of recommended items; on the other hand, the traditional recommendation method relies too much on the user's rating of the item To make predictions, it is impossible to completely avoid the user's rating bias and malicious rating data, resulting in deviations between the recommended items and the items that the user is interested in

Images