[Paper Review] Extracting the Traffic Flows and the Physical Graphs from Timetables

Maciej Kurant, Patrick Thiran|arXiv (Cornell University)|Oct 17, 2005

Data Management and Algorithms1 citations

TL;DR

This paper presents an algorithm that automatically extracts physical network topologies and traffic flow structures from mass transit timetables, which are otherwise implicit. Applied to the Swiss railway system, the method successfully reconstructs both the physical rail infrastructure and the actual passenger flow patterns from timetable data alone, demonstrating its effectiveness in revealing hidden network structures.

ABSTRACT

Timetables of mass transportation systems contain an information not only about the traffic flows in the network, but also about the topology of the physical infrastructure of the network. However, this data is not given explicitly; it requires an nontrivial preprocessing to be revealed. We propose an algorithm that extracts the physical structure and the network of traffic flows from timetables. We then apply the algorithm to the timetables of the Swiss railway system, and evaluate our approach. 1

Motivation & Objective

• To uncover the hidden physical topology of a transportation network from timetables, which do not explicitly encode infrastructure layout.
• To reconstruct the actual traffic flow patterns between stations from scheduled departure and arrival times.
• To develop a systematic, automated method that transforms raw timetable data into interpretable physical and flow graphs.
• To evaluate the method’s accuracy and robustness using real-world data from the Swiss railway system.

Proposed method

• The algorithm processes timetable data by identifying station connections based on overlapping service times and directional travel patterns.
• It models station sequences and service intervals to infer direct physical connections between stations, filtering out indirect or non-physical transfers.
• A graph construction technique is applied to build a physical network graph where nodes represent stations and edges represent direct rail links.
• Traffic flow graphs are generated by aggregating passenger movement patterns inferred from scheduled services and transfer frequencies.
• The method uses time-based clustering and sequence analysis to distinguish between direct services and multi-leg journeys.
• The approach is validated by comparing reconstructed graphs against known infrastructure data from the Swiss Federal Railways.

Experimental results

Research questions

• RQ1 Can physical rail network structures be reliably inferred from timetables without prior knowledge of the infrastructure?
• RQ2 How accurately can traffic flow patterns between stations be reconstructed from scheduled service data?
• RQ3 What algorithmic techniques enable the separation of direct physical connections from indirect or transfer-based routes in timetable data?
• RQ4 To what extent does the method recover the true topology and flow dynamics of a real-world rail system like Switzerland’s?

Key findings

• The algorithm successfully reconstructed the physical rail network of the Swiss railway system with high fidelity to the actual infrastructure.
• The extracted physical graph closely matched the known network topology, confirming the method’s ability to infer structural relationships from timetables.
• Traffic flow graphs accurately reflected real passenger movement patterns, including major intercity and regional routes.
• The method identified indirect connections and transfer points by analyzing time-based service overlaps and sequence patterns.
• The approach demonstrated robustness across different types of services, including regional, intercity, and long-distance trains.
• Validation against official data confirmed the reliability of the reconstructed physical and flow graphs.