Abstract

Link prediction is one of the most essential and significant research issues in network science, which aims at finding unknown, missing or future links in complex networks. Numerous similarity-based algorithms have been widely applied due to low computational cost and high prediction accuracy. To achieve a good balance between prediction accuracy and computational complexity, we design a novel link prediction algorithm to predict potential links based on Communication Capabilities and Local Paths (CCLP). The core idea of the proposed algorithm is that the similarity of node pairs is closely bound up with the amount of resources transmitted between them. In this algorithm, we introduce communicability network matrix to calculate the resource transmission capability of nodes, and then integrate it with local paths to measure the amount of resources transferred between nodes. We conduct several groups of comparative experiments on 12 real-world networks to validate the effectiveness of the proposed algorithm. Experimental results demonstrate that CCLP has achieved better performance than four classical similarity indices and five popular algorithms.

Graphic abstract

中文翻译：

---

基于通信能力和本地路径的复杂网络链路预测

摘要

链接预测是网络科学中最重要和最重要的研究问题之一，旨在发现复杂网络中未知、缺失或未来的链接。由于计算成本低、预测精度高，许多基于相似性的算法得到了广泛的应用。为了在预测精度和计算复杂度之间取得良好的平衡，我们设计了一种新的链路预测算法来预测基于通信能力和局部路径（CCLP）的潜在链路。该算法的核心思想是节点对的相似性与它们之间传输的资源量密切相关。在该算法中，我们引入通信网络矩阵来计算节点的资源传输能力，然后将其与本地路径集成以测量节点之间传输的资源量。我们在 12 个真实世界的网络上进行了几组比较实验，以验证所提出算法的有效性。实验结果表明，CCLP 比四种经典相似度指标和五种流行算法取得了更好的性能。

图形摘要