Recommended method, device, equipment, storage medium and program product

CN121579796BActive Publication Date: 2026-06-09TIANJIN UNIV

Patent Information

Authority / Receiving Office: CN · China
Patent Type: Patents(China)
Current Assignee / Owner: TIANJIN UNIV
Filing Date: 2026-01-29
Publication Date: 2026-06-09

AI Technical Summary

Technical Problem

Existing recommendation methods ignore the causal relationships between user behaviors, leading to false associations and inaccurate recommendations. Furthermore, when the number of auxiliary behaviors exceeds the number of target behaviors, the model becomes overly reliant on auxiliary behavior signals, masking the user's true intent and resulting in a decline in recommendation quality.

Method used

By processing the target behavior matrix and auxiliary behavior matrix through a graph convolutional neural network, conditional probabilities and marginal probabilities are determined, the causal structure is characterized, and the interference of auxiliary behaviors on the recommendation results is eliminated. The graph convolutional neural network is trained with a Bayesian personalized ranking algorithm to improve the recommendation accuracy.

Benefits of technology

In scenarios with excessive and unevenly distributed auxiliary behaviors, the goal is to maintain the accuracy and robustness of recommendations, avoid bias caused by spurious relevance, and improve recommendation quality and interpretability.

✦ Generated by Eureka AI based on patent content.

Smart Images

Figure CN121579796B_ABST

Patent Text Reader

Abstract

This application provides a recommendation method, apparatus, device, storage medium, and program product, which can be applied to the field of information technology. The method includes: determining a target behavior matrix and multiple auxiliary behavior matrices corresponding to multiple auxiliary behavior types based on user interaction behavior with multiple products; processing reference auxiliary behavior matrices in the target behavior matrix and multiple auxiliary behavior matrices using a graph convolutional neural network to determine conditional probabilities, whereby the conditional probabilities characterize whether a user is recommended a product given that the user performs a reference auxiliary behavior and a purchase behavior; determining marginal probabilities of reference auxiliary behaviors based on the reference auxiliary behavior matrices, whereby the marginal probabilities characterize the contribution of performing reference auxiliary behaviors to the recommended products; determining the recommendation degree for each product based on the conditional probabilities and marginal probabilities corresponding to the multiple auxiliary behavior types, and pushing target product information determined based on the recommendation degree to the user.

Need to check novelty before this filing date? Find Prior Art