RouteKG: A knowledge graph-based framework for route prediction on road networks

TL;DR

RouteKG reframes short-term route prediction as a knowledge graph completion task to explicitly model road-network spatial relations, including moving directions, rather than relying solely on sequential pattern learning. The framework combines a data preprocessing pipeline, a Knowledge Graph Module with four relation types (ConnectBy, ConsistentWith, DistanceTo, DirectionTo), an $n$-ary Spanning Route generator, and a Rank Refinement module to produce and rerank top-$K$ route predictions. Empirical results on Chengdu and Shanghai taxi data show RouteKG consistently outperforms baselines across NoGoal, GoalD, and Goal scenarios, with notable gains when goal information is available, and enables accurate link-level traffic-flow estimation. The approach offers real-time applicability (sub-second per-10k requests) and demonstrates the potential of integrating knowledge graphs with routing, while suggesting directions for extending to dynamic contexts and longer-horizon predictions.

Abstract

Short-term route prediction on road networks allows us to anticipate the future trajectories of road users, enabling various applications ranging from dynamic traffic control to personalized navigation. Despite recent advances in this area, existing methods focus primarily on learning sequential transition patterns, neglecting the inherent spatial relations in road networks that can affect human routing decisions. To fill this gap, this paper introduces RouteKG, a novel Knowledge Graph-based framework for route prediction. Specifically, we construct a Knowledge Graph on the road network to encode spatial relations, especially moving directions that are crucial for human navigation. Moreover, an n-ary tree-based algorithm is introduced to efficiently generate top-K routes in batch mode, enhancing computational efficiency. To further optimize prediction performance, a rank refinement module is incorporated to fine-tune candidate route rankings. The model performance is evaluated using two real-world vehicle trajectory datasets from two Chinese cities under various practical scenarios. The results demonstrate a significant improvement in accuracy over the baseline methods. We further validate the proposed method by utilizing the pre-trained model as a simulator for real-time traffic flow estimation at the link level. RouteKG has great potential to transform vehicle navigation, traffic management, and a variety of intelligent transportation tasks, playing a crucial role in advancing the core foundation of intelligent and connected urban systems. The source codes of RouteKG are available at https://github.com/YihongT/RouteKG.

Figures (10)

• Figure 1: The workflow of RouteKG.
• Figure 2: An illustration of discretized directions with examples of inter- and intra-edge directions.
• Figure 3: An illustration of many-to-many spatial relations between sampled head and tail entities. Each relation captures specific patterns between entity pairs: (a) ConnectBy: adjacency in the road network via a shared node; (b) ConsistentWith: co-occurrence in observed routes; (c) DistanceTo: spatial or network distance; (d) DirectionTo: navigational direction.
• Figure 4: An illustration of knowledge graph completion for future route prediction.
• Figure 5: A demonstration of the Spanning Route algorithm.

Theorems & Definitions (2)

• Definition 1: Road Network $\mathbf{G}$
• Definition 2: Route $x$