Paper 2021/1356

Structural Mutual Information and Its Application

Youliang Tian, Zhiying Zhang, Jinbo Xiong, and Jianfeng Ma


Shannon mutual information is an effective method to analyze the information interaction in a point-to-point communication system. However, it cannot solve the problem of channel capacity in graph structure communication system. This problem make it impossible to use traditional mutual information (TMI) to detect the real information and to measure the information embedded in the graph structure. Therefore, measuring the interaction of graph structure and the degree of privacy leakage has become an emerging and challenging issue to be considered. To solve this issue, we propose a novel structural mutual information (SMI) theory based on structure entropy model and the Shannon mutual information theorem, following by the algorithms for solving SMI. The SMI is used to detect the real network structure and measure the degree of private data leakage in the graph structure. Our work expands the channel capacity of Shannon’s second theorem in graph structure, discusses the correlation properties between SMI and TMI, and concludes that SMI satisfies some basic properties, including symmetry, non-negativity, and so on. Finally, theoretical analysis and example demonstration show that the SMI theory is more effective than the traditional privacy measurement methods to measure the information amount embedded in the graph structure and the overall degree of privacy leakage. It provides feasible theoretical support for the privacy protection technology in the graph structure.

Available format(s)
Publication info
Preprint. Minor revision.
Structural mutual information (SMI)mutual informationstructure entropyprivacy measurementprivacy leakage.
Contact author(s)
13595150727 @ 163 com
2021-10-12: received
Short URL
Creative Commons Attribution


      author = {Youliang Tian and Zhiying Zhang and Jinbo Xiong and Jianfeng Ma},
      title = {Structural Mutual Information and Its Application},
      howpublished = {Cryptology ePrint Archive, Paper 2021/1356},
      year = {2021},
      note = {\url{}},
      url = {}
Note: In order to protect the privacy of readers, does not use cookies or embedded third party content.