Social robot detection method based on directed relation graph comparative learning

Through the method based on directed relationship graph comparison learning, combined with attention mechanism and graph comparison learning, the problem of insufficient detection accuracy of existing social robots is solved, significantly improving the recognition accuracy and enhancing the robustness of the detector.

CN119312166AActive Publication Date: 2025-01-14HANGZHOU DIANZI UNIV
View PDF 14 Cites 2 Cited by

Patent Information

Application Number
CN202411503435.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-25
Publication Date
2025-01-14
Estimated Expiration
2044-10-25

AI Technical Summary

Technical Problem

The accuracy of existing social robot detection methods is insufficient, making it difficult to identify automated robot accounts that mimic real users' behaviors. Especially because these robots simulate the characteristics and relationships of real users through feature operations, resulting in poor detection accuracy.

Method used

Using a method based on directed relationship graph comparison learning, the directed relationship graph is constructed by obtaining the initial feature vector of user nodes, combining attention mechanism and graph comparison learning, and using graph comparison view to enhance the similarity between users of the same category and different categories. Differences between users improve detection accuracy.

Benefits of technology

It significantly improves the recognition accuracy of robot accounts on social media platforms, enhances the robustness of the detector, and can more effectively distinguish between real users and social robots.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119312166A_ABST
    Figure CN119312166A_ABST
Patent Text Reader

Abstract

The invention discloses a social robot detection method based on directed relation graph comparative learning, and the method comprises the steps: taking each user as a user node, obtaining the personal attribute information and the personal text information of each user node, splicing the personal attribute information and the personal text information, and converting the spliced information into a feature vector; constructing a user directed relation graph, learning features of adjacent users under different relations by using a directed relation graph convolutional neural network, and performing relation feature fusion based on an attention mechanism; a comparison view of the directed relation graph is constructed from the two aspects of attribute features and graph topological structures in combination with direction information, and then training and prediction are carried out. According to the method, the accuracy of social network platform robot account identification is improved through the directed relation graph attention mechanism and directed graph comparative learning, and the directed relation graph is combined with the attention mechanism to effectively utilize the direction information of the social network platform relationship and the importance information of the relationship between the users in real life.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Multi-view hyperbolic-hyperbolic graph representation learning method

    CN115906946A

  • Social robot detection system and method based on ternary graph convolutional neural network

    CN116186636A

  • Data-driven block chain anomaly detection method and terminal

    CN116361734A

  • Social robot detection method, device and equipment and storage medium

    CN116720009A

  • Social robot detection method and device based on multi-dimensional feature fusion and residual image neural network

    CN116738331A

Cited By

  • Social recommendation method and system based on structure and semantic double-view comparison and medium

    CN120277278A

  • Social robot detection method based on heterogeneous edge decoupling and balanced label expansion

    CN122335283A