Scholarlink

<p>Community deception in social networks is a strategy used to address privacy breaches caused by community detection algorithms. It involves intentionally disrupting the network structure to deceive or mislead community detection algorithms and protect the privacy of individuals or groups. Recently, the study of community deception has gained increasing attention and achieves success to some extent in this area. However, despite these advancements, most studies still suffer from three major shortcomings. First, they design the community deception metrics in accordance with the definition of community, which fails to evaluate the effectiveness of community deception directly to guide the optimization process. Second, they change the structure of the original network by deleting or adding redundant nodes and edges to it, making the changes easy to be discovered by some basic network properties. Third, in terms of the widely used genetic algorithm (GA), as the scale of the complex network increases, the search space for the GA also expands, posing a greater challenge for the GA to find the optimal solution. To fill these critical voids, we first propose to exploit normalized mutual information (NMI) to evaluate the effectiveness of community deception and use it as a fitness function to guide the community deception method in searching for the optimal solution. Then, we design a novel method that disrupts the structure of the original network by adding and deleting edges, ensuring an equal number of additions and deletions, with the aim to enhance concealment and address the issue of inadequate deception. Finally, we propose a Co-evolutionary genetic algorithm that utilizes an Elite Population, CGAEP for brevity, to improve the effectiveness and efficiency of the search for the optimal solution in community deception. Extensive experiments on seven real-world networks demonstrate that CGAEP achieves consistent advantages over the state-of-the-art methods in terms of community deception’s effectiveness and efficiency.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Global community deception via a cooperative evolutionary genetic algorithm based on an elite population

  • 期刊论文

  • 发表日期：

    2026年7月2日

查看全文
  • Guixiang Zhu,
  • Lei Chen,
  • Haobin Cao,
  • Fumin Ma,
  • Shuxin Yang,
  • Baizhen Chen

摘要

Community deception in social networks is a strategy used to address privacy breaches caused by community detection algorithms. It involves intentionally disrupting the network structure to deceive or mislead community detection algorithms and protect the privacy of individuals or groups. Recently, the study of community deception has gained increasing attention and achieves success to some extent in this area. However, despite these advancements, most studies still suffer from three major shortcomings. First, they design the community deception metrics in accordance with the definition of community, which fails to evaluate the effectiveness of community deception directly to guide the optimization process. Second, they change the structure of the original network by deleting or adding redundant nodes and edges to it, making the changes easy to be discovered by some basic network properties. Third, in terms of the widely used genetic algorithm (GA), as the scale of the complex network increases, the search space for the GA also expands, posing a greater challenge for the GA to find the optimal solution. To fill these critical voids, we first propose to exploit normalized mutual information (NMI) to evaluate the effectiveness of community deception and use it as a fitness function to guide the community deception method in searching for the optimal solution. Then, we design a novel method that disrupts the structure of the original network by adding and deleting edges, ensuring an equal number of additions and deletions, with the aim to enhance concealment and address the issue of inadequate deception. Finally, we propose a Co-evolutionary genetic algorithm that utilizes an Elite Population, CGAEP for brevity, to improve the effectiveness and efficiency of the search for the optimal solution in community deception. Extensive experiments on seven real-world networks demonstrate that CGAEP achieves consistent advantages over the state-of-the-art methods in terms of community deception’s effectiveness and efficiency.