当前位置: X-MOL 学术Social Networks › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Link prediction in growing networks with aging
Social Networks ( IF 4.144 ) Pub Date : 2020-11-12 , DOI: 10.1016/j.socnet.2020.11.001
Li Zou , Chao Wang , An Zeng , Ying Fan , Zengru Di

The link prediction problem aims to predict new links for future, or missing links or unobserved links in complex networks. Traditional link prediction methods are mostly concentrated on static networks. In this paper, we mainly explore link prediction problems in growing networks. We propose a series of time-sliced metrics to estimate the likelihood of the existence of missing links between two nodes for evolving networks based on traditional link prediction indices. We found that these proposed metrics outperform existing metrics for growing networks with time decay factor, especially when the decay factors are small. Besides, to improve prediction efficiency and practicability, we propose the function expressions for optimal slice number and decay factor for real-world networks. The formula enables us to estimate the aging speed of real growing networks, resulting in accurate and fast prediction of missing links in growing networks.



中文翻译:

老化网络中的链接预测

链路预测问题旨在预测将来的新链路,或者预测复杂网络中的链路丢失或未观察到的链路。传统的链路预测方法主要集中在静态网络上。在本文中,我们主要探讨成长网络中的链接预测问题。我们提出了一系列按时间划分的指标,以基于传统的链路预测指标来估计两个节点之间存在缺失链路的可能性,以用于演进的网络。我们发现,对于具有时间衰减因子的成长网络,这些拟议指标优于现有指标,尤其是在衰减因子较小时。此外,为了提高预测效率和实用性,我们提出了现实网络中最佳切片数和衰减因子的函数表达式。

更新日期:2020-11-12
down
wechat
bug