A sequential-path tree-based centrality for identifying influential spreaders in temporal networks

The problem of identifying influential spreaders in temporal networks has attracted extensive attention in recent years. Existing studies have proposed various centrality measures for quantifying the influence of nodes based on the structures of temporal networks, such as temporal degree centrality and temporal closeness centrality. However, most existing methods only take into account a single feature of nodes, while ignoring other temporal features (e.g., the propagation time, or the path length from an infected node to a destination node). In this paper, we propose a new centrality measure, namely as sequential-path tree-based centrality (SPT-C) which takes into account three different temporal features based on a new representation structure of temporal networks (i.e., sequential-path tree). The three temporal features include propagation time which measures the time that an infection propagates from an infected node to another, hop count which denotes the number of intermediate nodes that an infection propagates from an infected node to another, and reachable paths which represent the number of different time-respecting paths from an infected node to another. The evaluation experiments on 12 real-world temporal networks show that the effectiveness of our SPT-C in identifying influential spreaders is superior to other baseline measures.