Intelligent recommendation method and system based on user preference correction

Abstract

The invention discloses an intelligent recommendation system based on user preference correction. Comprising the steps of performing user preference analysis on a user according to evaluation of the user on a product; designing a score mapping algorithm, and subjecting the users with different evaluation preferences to average distribution processing of evaluation; projecting the evaluation conditions of the users with different evaluation preferences to a fixed score interval in a scoring form; better matching the scoring preference of the user and the comprehensive evaluation of the product, and improving reasonable matching of the user and the product. And then the processed scoring features and the user features jointly form a user portrait matrix, so that personalized recommendation of'user-product 'is realized. According to the invention, the problem of user-product bidirectional matching in industry is solved. The invention can be applied to industrial scenes of recommending users for products and recommending products for the users.

Description

technical field [0001] The invention relates to an intelligent recommendation system and method based on user preference correction, and belongs to the technical field of intelligent recommendation. Background technique [0002] In recent years, with the development of information technology and the Internet, information has shown a geometric explosive growth. The problem people face has changed from lack of information to how to filter useful information from massive data. This is an era when life is affected by recommendation systems everywhere. Based on massive data, recommendation systems provide targeted product information recommendations for users' different preferences based on data generated by user behavior. In the industry, the recommendation system has gradually become one of the research objects. For users, the recommendation system solves the problem of how to efficiently obtain interesting information under the condition of "information overload". For the com...

AI Technical Summary

Problems solved by technology

[0004] The present invention starts from the perspective that different users have their own personalized product evaluation preferences, solves the impact of user scoring and evaluation preferences on the recommendation system, and avoids recommending low-scoring products to habitual poor-evaluating users or high-scoring products that cannot be accepted. Recommend to habitual praise users with higher average ratings. For example, user 1 is a habitual negative rating user with an average rating of 1.8, then the product with a predicted score of 2 is likely to be included in the recommendation list, and this comprehensive evaluation A product with a score of 2 is only about 1.2 in the user's scoring habits, and it will not be able to meet the needs of the user; user 2 is a habitual praise user with an average score of 4.5, and a product with a score of 4 will not Included in the recommendation category, but this product with a comprehensive evaluation of 4 points is already at a level of 4.8 or higher in the user's scoring habit, so the product cannot get this customer and cannot meet the seller's needs

Images