Application of Graph Learning With Multivariate Relational Representation Matrix in Vehicular Social Networks

The essence of connection in vehicle network is the social relationship between people, and thus Vehicular Social Networks (VSNs), characterized by social aspects and features, can be formed. The information collected by VSNs can be used for context prediction of autonomous vehicles. Multivariate relations are common in square connected relations caused by geographic characteristics in VSNs. They can effectively reflect the high-order structural features of the network dataset. It is necessary to exploit the multivariate relations of VSNs to improve the performance of context prediction. However, The representation of entity-relationes in the network often adopts a binary form, and the existing graph learning methods rely on the neighborhood information of nodes to achieve the aggregation or diffusion of information. Using this to represent multivariate relations will result in partial omissions or even complete loss of valuable information, which ultimately affects the learning effect of learning methods. In order to better understand the social behavior of the VSNs, this paper uses the network motif to implement the representation of the multivariate relations in the network, and proposes the graPh learnIng with moTif mAtrix (PITA) method. This method can be used as a preprocessing step for the measurement strategy of the relations in VSNs and the graph learning, which can mine the information in VSNs and improve the accuracy of the original graph learning method by the multivariate relation information. The authors performed experiments on 6 network datasets. The experimental results show that in the node classification task, the baseline method modified by the PITA method has a higher classification accuracy than the original method.

• Record URL:
  http://dx.doi.org/10.1109/TITS.2022.3224326
• Availability:
  • Find a library where document is available. Order URL: http://worldcat.org/oclc/41297384
• Supplemental Notes:
  • Copyright © 2023, IEEE.
• Authors:
  • Wan, Liangtian
  • Li, Xiaona
  • Xu, Jin
  • Sun, Lu
  • Wang, Xianpeng
  • Liu, Kaihui
• Publication Date: 2023-3

Language

• English

Media Info

• Media Type: Web
• Features: References;
• Pagination: pp 2789-2799
• Serial:
  • IEEE Transactions on Intelligent Transportation Systems
  • Volume: 24
  • Issue Number: 3
  • Publisher: Institute of Electrical and Electronics Engineers (IEEE)
  • ISSN: 1524-9050
  • Serial URL: http://ieeexplore.ieee.org/xpl/RecentIssue.jsp?punumber=6979

Subject/Index Terms

• TRT Terms: Autonomous vehicles; Connected vehicles; Machine learning; Social factors; Vehicular ad hoc networks
• Subject Areas: Data and Information Technology; Highways; Society; Vehicles and Equipment;

Filing Info

• Accession Number: 01888966
• Record Type: Publication
• Files: TRIS
• Created Date: Jul 26 2023 3:59PM