基于无限制路径、步行和循环的网络中心性指标与标准中心性指标比较
社会与信息网络
2026-02-09 v2
摘要
传统的接近性和介于中心性度量依赖于节点之间的最短路径。许多标准度量未能准确反映节点中心性和网络流的物理或概率特征，常忽视循环和循环扩散等过程。本文提出新的度量基于我们的影响扩散模型。这些概率度量考虑网络中所有可行路径、步行和循环。我们定义入中心性以评估节点作为影响目标的中心性，定义出中心性以衡量其作为影响源的作用。我们通过分析节点排序、散点图以及计算皮尔逊相关系数和斯佩尔曼秩相关系数来比较这些度量与标准度量。我们的发现表明，基于影响扩散模型定义的介于中心性强调了备用路径的重要性，同时保持与标准介于中心性的相似性。
引用
@article{arxiv.2512.11585,
title = {Network Centrality Metrics Based on Unrestricted Paths, Walks and Cycles Compared to Standard Centrality Metrics},
author = {Juuso Luhtala and Vesa Kuikka and Kimmo K. Kaski},
journal= {arXiv preprint arXiv:2512.11585},
year = {2026}
}
备注
The manuscript has been substantially shortened from 37 pages to 27 pages. We have extended the analysis to include Spearman's rank correlation coefficients alongside Pearson correlations, which improves the robustness of the comparison, particularly for non-linear relationships and rank-based assessments in heavy-tailed networks. The readability of some figures has been improved