Scientific and technological resource dynamic collaborative filtering recommendation method based on user group preferences

Abstract

The embodiment of the invention provides a scientific and technological resource dynamic collaborative filtering recommendation method based on user group preferences. The method comprises the following steps: calculating the similarity among users according to the label similarity among the users and the score similarity of the users on scientific and technological resources; clustering the users, and constructing a category preference matrix of the science and technology resources; calculating the category preference similarity of the user to the target science and technology resources and other science and technology resources, and generating a candidate neighbor list of the target science and technology resources; calculating score similarity between each science and technology resource and a plurality of previous science and technology resources, and generating a science and technology resource nearest neighbor list; taking the science and technology resources which are not scoredby the user as to-be-predicted science and technology resources, and calculating prediction scores of the to-be-predicted science and technology resources. In this way, the score data can be predicted by analyzing the score of the similar user group of the target user, and the generated recommendation result is highly consistent with the interest matching degree of the target user.

Description

technical field [0001] Embodiments of the present invention generally relate to the field of computer technology, and more specifically, relate to a dynamic collaborative filtering recommendation method for scientific and technological resources based on preferences of user groups. Background technique [0002] Collaborative filtering algorithm is the most widely used core algorithm in recommendation system. Collaborative filtering is also called social filtering (Social Filtering). This concept was first proposed in Goldber's research report in 1992. Goldberg et al. applied the Tapestry system to filter out useful electronic letters for users, but required users to mark The information you do not want to see reflects the interaction between the system and the user. So far, collaborative filtering technology has been successfully applied to various recommendation systems. The core idea is: the machine itself lacks emotional elements, which causes defects in the content it s...

AI Technical Summary

Problems solved by technology

However, there is a problem in traditional collaborative filtering algorithms. They only consider the similarity of ratings between users or scientific and technological resources, and do not take into account changes in user interests. User interests are not static and change over time. Now The content that users are interested in in one stage may not be still interesting in the next stage

Images