Triangle formation represents a fundamental mechanism in network evolution. We perform a comprehensive analysis of triangle closure patterns across six real-world communication networks, examining both temporal dynamics and community-based formation behaviors. Temporally, networks representing direct communications exhibit tightly clustered interarrival times indicative of sustained communication, whereas networks representing indirect communications display widely varying durations driven by bursty closure events. From a community perspective, direct communication networks with strong within-community closure demonstrate stable behavior across persistence thresholds, while those with indirect communication and weaker closure rates show substantial variability, suggesting less cohesive community boundaries. We also present a ridge regression-based triangle prediction framework that performs on par with state-of-the-art methods while requiring significantly less information about networks. Our findings contribute to advancing the understanding of network evolution and providing tools for network analysis.

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

Dynamics of Triadic Closure in Complex Networks

  • Aashish Pandey,
  • Arvind Prasadan,
  • Rich V. Field,
  • Jeremy D. Wendt,
  • Cynthia A. Phillips,
  • Sucheta Soundarajan,
  • Sanjukta Bhowmick

摘要

Triangle formation represents a fundamental mechanism in network evolution. We perform a comprehensive analysis of triangle closure patterns across six real-world communication networks, examining both temporal dynamics and community-based formation behaviors. Temporally, networks representing direct communications exhibit tightly clustered interarrival times indicative of sustained communication, whereas networks representing indirect communications display widely varying durations driven by bursty closure events. From a community perspective, direct communication networks with strong within-community closure demonstrate stable behavior across persistence thresholds, while those with indirect communication and weaker closure rates show substantial variability, suggesting less cohesive community boundaries. We also present a ridge regression-based triangle prediction framework that performs on par with state-of-the-art methods while requiring significantly less information about networks. Our findings contribute to advancing the understanding of network evolution and providing tools for network analysis.