Temporal path-based centrality measure for event-based road networks

Temporal networks capture systems in which interactions evolve over time. In transportation networks, this temporal structure governs the propagation of congestion and determines how junctions influence network-wide traffic flow. Existing temporal centrality measures primarily rely on temporal reachability or shortest paths, treating all feasible time-respecting paths equivalently once temporal ordering is satisfied while overlooking differences in travel delay, congestion and structural efficiency. To address this limitation, we propose Temporal Path-Based Centrality (TPBC), a centrality measure that evaluates node importance using three components: Relative Temporal Delay (RTD), Congestion Delay (CD) and Relative Hop (RH). Rather than relying on a single optimal path, TPBC aggregates contributions from all feasible time-respecting paths, weighting each according to its delay, congestion and structural efficiency. GPS trajectories from the T-Drive dataset are aligned to the road network using Hidden Markov Model (HMM)-based map matching to construct an event-based temporal network. To ensure computational tractability, path enumeration is bounded using practical limits on path length and the number of feasible paths. Experiments on the Chaoyang district road network across seven days demonstrate that TPBC exhibits the strongest correlation with observed traffic volume while capturing aspects of node importance not reflected by purely structural or instantaneous road-condition measures. TPBC identifies influential junctions predominantly on residential and lower-order roads rather than major arterial infrastructure. Node-removal experiments show that TPBC-selected junctions produce substantially greater reductions in time-respecting connectivity and network efficiency than random removal across all seven days. Sensitivity analyses confirm that these findings remain robust to variations in both the path-enumeration bounds and the relative weighting of delay, congestion and structural efficiency. These results establish TPBC as a robust, data-driven framework for identifying critical nodes in temporal transportation networks, with applications in traffic management, infrastructure planning and urban resilience.