RSM recast, based on direct RDF/OWL modelling.
RSM stands for Rail System Model. RSM is a UIC IRS (International Railway Standard), first released in 2016 as RailTopoModel (RTM): see https://rsm.uic.org.
EUPL 1.2
RTM 1.0, RTM 1.1, RSM 1.2 and its extensions, all published by UIC, follow principles generally observed in conceptual modelling, using UML 2.56 as a formal language. RSM 1.2 model was successfully transformed into an OWL ontology, using the Ontorail toolset developed by UIC.
However, the expressiveness of OWL differs from UML class diagrams. OWL offers possibilities to make RSM both more compact and more expressive, while remaining compatible with former, UML-based versions. The present repository summarizes the re-casting efforts.
- Backward compatibility ("easy" automated transformation of data provided under previous versions of the model).
- Simplification (not: dumbing down).
- Separation into small vocabularies (high-cohesion, low-dependency principle) to facilitate ontology management.
- Improvements:
- about topology:
- navigability as a transitive property,
- internal navigability in non-linear elements such as "operational points" (yards, stations),
- more flexible composition of net elements.
- about topology:
- Systematic, explicit usage of well-established external vocabularies, where relevant: SOSA/SSN, GeoSPARQL, systems of units ontologies, w3c time...
- Ability to determine paths under constraints using SPARQL and inference engines, even though bespoke code may be recommended for computing efficiency.
The process currently considers:
- the RINF use case, with priority to topology (expressed at micro level, or track level) and geographic referencing.
- the System Pillar requirements regarding rolling stock modelling (typology)
- FP5-TRANSF4M-R data requirements regarding the description of the "last mile of infrastructure" and of rolling stock defects.
- Any relevant FPx-MOTIONAL works.
However the design, as previously with RSM, emphasizes generality and usability, while avoiding ad-hoc solutions.
Other important concepts may come from other pieces of EU Law, such as TAF TSI, or from ongoing EU projects.
- Protégé desktop 5.x for RDF edition, checking, and saving to different formats.
- EasyRdf for format conversion (between RDF/XML and Turtle, mostly).
- OOPS! pitfall scanner, in addition to Protégé, for ontology checking.
- Graphics:
- Sparx Enterprise Architect (v. 16 or later) for UML diagrams, possibly using the built-in ODM UML profile.
- Visual Paradigm also for UML diagrams (most illustrations come from Visual Paradigm).
- Draw.io for hand-drawn graphics.
- Some diagrams (in the Wiki) were generated from extended markdown using Mermaid scripts, or
- from ttl files using OntoMermaid, a Python program returning a Mermaid script.
- from IFC Step files, using OpenIFCViewer. As Sparx Enterprise Architect and Visual Paradigm are paid applications, copies of the diagrams are also provided in PNG.
- Using a schematic track plan made in draw.io to generate a sRSM file in RDF Turtle.
- Using an OpenStreetMap query output to generate a sRSM file.
See the Flask folder and launch __init__.py. You'll be able to process files with your usual web navigator, via a local server (127.0.0.1:8000).
- Sample data sets:
- based on fictive or real networks, or
- based on OpenStreetMap, using Overpass Turbo: these data are from www.openstreetmap.org. The data are made available under ODbL (see this page).
- from actual sample sets (e.g. Scheibenberg test line).
RDF, RDF-star, OWL, SHACL (for constraints), SPARQL (for queries), SPARQL-star, Python; some SWI Prolog (for demo purposes).
The Wiki documents the design process, esp. main design choices. This is an integrated wiki, also accessible via the GitHub menu bar.
Only contains diagrams (class diagrams for instance) and illustrations for use by the wiki and in other repositories. Full documentation is in the wiki.