Efficient Influential Nodes Tracking via Link Prediction in Evolving Networks

Kexin Zhang, Taotao Cai, Zhaoyu Liu, Shuang Teng, Yi Wang, Yu Chen, Ji Zhang

Abstract: Influence maximization (IM), which aims to identify the most influential k nodes in a network, is fundamental to numerous applications, including viral marketing and recommendation systems. This topic has garnered significant scholarly attention. However, most existing research addresses the IM problem in static networks, neglecting the dynamic and continually evolving nature of social networks. In this article, we introduce a novel problem: influential nodes tracking in future networks (INTFN). The INTFN problem aims to quickly find the most influential k nodes in networks over upcoming time intervals. We formally define the INTFN problem and prove its NP-hardness. To address this challenge, we propose a comprehensive solution that predicts the future structure of social networks using a carefully selected link prediction technique. Subsequently, we identify the most influential k nodes in these future networks by employing classic IM algorithms. Additionally, we design a dictionary structure and propose the compressed subgraphs-based influential nodes tracking (CSINT) algorithm to enhance the efficiency of our solution. Extensive experiments on four real-world datasets demonstrate the effectiveness and efficiency of the proposed CSINT algorithm.

External IDs:dblp:journals/iotj/ZhangCLTWCZ25