Group user-oriented information recommendation method and computer readable storage medium

Abstract

The invention discloses a group user-oriented information recommendation method and a computer readable storage medium. The method comprises the steps of obtaining classification information of recommended articles; performing topic contribution extraction on the classification information of the recommended articles, and extracting an implicit topic preference feature matrix in combination with analysis of activity logs of the user; projecting the article and the user log record to a theme preference feature matrix, and calculating the attraction degree of the article to the user group; and obtaining an article rank according to the attraction degree of the article to the user group and recommending the article to the user. According to the method, the TF-IDF is adopted to extract the theme information of the articles, and the situation that each article belongs to multiple articles at the same time instead of being simply classified into one category is considered, so that most scenerequirements are met; the problem that a certain article is frequently recommended due to malicious refreshing and clicking is effectively avoided, and malicious interference of a user on a recommendation model is reduced; and the recommendation result set contains various different types of information, so that over-fitting is avoided.

Description

technical field [0001] The present invention relates to the technical field of information recommendation, in particular to an information recommendation method for group users and a computer-readable storage medium. Background technique [0002] The recommendation system is widely used in e-commerce, advertising marketing and other major network platforms. It can recommend popular products or information based on user behavior, such as popular movies and popular games. At present, technical means such as collaborative filtering, matrix decomposition, and regression model are usually used for the recommendation of precise users, but these means are mostly recommendation systems for specific users or a type of users; group-oriented recommendations are different from those of these recommendation systems. The point is that the objects of the recommendation service are different. The service object of the group recommendation system is no longer a specific user or a specific ty...

AI Technical Summary

Problems solved by technology

[0003] 1. It does not consider the different contributions of various categories of items;

[0004] 2. It does not consider the preference of the user group itself for different categories;

[0005] 3. There is no distinction between the distribution of the increase in frequency in the time direction, so it is impossible to distinguish the recent popularity from the long-term popularity;

[0006] 4. It is sensitive to users' malicious clicks and collections, and the recommendation model is easily affected by a large number of malicious clicks;

[0007] 5. The diversity of recommendation types is not considered. The more recommendations, the more clicks you get, and the more clicks, the higher the recommendation degree, and it is easy to fall into the preference for certain types of overfitting

Images