DTInsight: A Tool for Explicit, Interactive, and Continuous Digital Twin Reporting

DTInsight is a systematic and automated tool for producing continuous reporting for Digital Twins (DTs). It generates an interactive conceptual architecture visualization—called a DT constellation—from an ontological description of the DT, and integrates the output into a CI/CD-driven reporting page.

Key Features

DTInsight is built around three pillars of DT reporting:

Pillar	What it does
Explicit Reporting	Model the DT using the 21 characteristics of the DT Description Framework (DTDF) in the Ontology Modeling Language (OML), providing a systematic and consistent description.
Interactive Reporting	Explore a DT constellation: a conceptual architecture showing data flow between Physical Twin components (Sensors, System, Environment, Operator, Machine) and DT capability layers (Models/Data, Enablers, Services). Supports real-time sensor data and 3D visualization.
Continuous Reporting	Auto-generate an up-to-date reporting page (characteristics table, architecture screenshot, interactive web embed) from a CI/CD pipeline on every commit.

YouTube Overview

Architecture & Technology Stack

DTInsight is built with the Godot Engine (v4.5, .NET/C#) and uses GDScript for the UI and visualization logic.

DTInsight/
├── MainScene/          # Main scene controller, CI/CD TCP server, settings
├── DTContainer/        # DT & PT constellation layout, link drawing, visual editing
├── GenericDisplay/     # Reusable DT component display, pop-up charts & scripts
├── Fuseki/             # SPARQL query catalogue, Fuseki server caller, data dump
├── RabbitMQ/           # RabbitMQ connection & real-time data handling (C#/.NET)
├── Camera/             # 2D camera controls (pan, zoom, keyboard)
├── Config/             # Global configuration (style, Fuseki, RabbitMQ, camera, etc.)
├── ControlPanel/       # Settings UI panel
├── Legends/            # Color legend for implementation status & timescale
├── Build/              # Pre-built exports (Linux, Web)
└── .github/workflows/  # GitHub Actions CI/CD for automated builds

Key integrations:

• Apache Jena Fuseki — serves the DTDF ontology as an RDF triple store; DTInsight queries it via SPARQL over HTTP.
• RabbitMQ — message broker for subscribing to real-time sensor data streams (via the .NET RabbitMQ client library).
• openCAESAR Rosetta — recommended IDE for editing the OML description model.
• GitHub Actions — CI/CD workflow that exports Linux and Web builds on every push to develop.

Getting Started

Note: You can view tutorial slides at https://1drv.ms/p/c/86984f2cdf02822a/IQDsOu-5YYM6QKA_Xj-bf8tYAROfy-sjk1sZlwdcaXVK_aY?e=CK9JFc

Prerequisites

• Godot Engine 4.5 (.NET / C# version) — download
• .NET 8 SDK — required for the RabbitMQ C# integration
• Apache Jena Fuseki — bundled via openCAESAR Rosetta, or standalone
• A DTDF-compatible OML ontology — see the DTDF repository
• (Optional) A running RabbitMQ instance publishing DT sensor data

Installation

1. Clone the repository:

   git clone https://github.com/oakeslabmtl/DTInsight.git

2. Open the project in Godot Engine 4.5 (.NET version).
3. Build the C# solution to restore the RabbitMQ.Client NuGet package:

   dotnet restore
   dotnet build

4. Run the project from the Godot editor or export it for your platform.

How to Use

Settings

Click the gear icon (bottom-left) to open the settings panel.

Loading an Ontology

Option A: Load from Fuseki (recommended for development)

1. Open your OML ontology project in openCAESAR Rosetta.

2. In the Gradle Tasks tab, start the Fuseki server with the startFuseki task and run the owlLoad task to push the ontology data.

   

   On Windows, you can also call these tasks with:

   gradlew.bat startFuseki
   

   and then

   gradlew.bat owlLoad
   

3. In DTInsight, click the "Call Fuseki" button:

   

Option B: Load from a dump file

If a dump was previously exported, select the file with the file picker and click "Load".

Creating a dump

After loading an ontology, enter a file path in the dump input box and click "Dump". The resulting file can also be inspected in a text editor to verify loaded data.

Navigating the DT Constellation

Once loaded, you see a top-down conceptual architecture of the DT, structured as:

• Left side (Physical Twin): Operator, Machine, System, System Environment, Sensors/Data Transmission
• Right side (Digital Twin): Models/Data → Enablers → Services

Directional arrows show inter-component dependencies and data flow.

Camera Controls

Action	QWERTY	AZERTY	Mouse
Move up	W	Z	Click + drag
Move down	S	S	Click + drag
Move left	A	Q	Click + drag
Move right	D	D	Click + drag
Zoom in	E	E	Scroll up
Zoom out	Q	A	Scroll down

Highlighting Relationships

Hover over a component to highlight it and its connected components, revealing data-flow paths. Click to lock the highlight.

Visual Editing

Enable Visual Editing mode from the control panel to:

• Add new components to any container via the + buttons
• Rename or edit descriptions by right-clicking a component
• Draw links by dragging from one component to another
• Delete links by enabling link deletion mode and clicking on a link

Real-Time Data (RabbitMQ)

If your DT publishes sensor data via RabbitMQ and your ontology includes the necessary linking metadata:

1. Toggle "Record data from RabbitMQ" in the settings.
2. Real-time values will appear on the corresponding DT components.
3. Click the pop-up button on a component to view a live chart of the last 100 messages (supports numerical and boolean values).

Note: If nothing appears after enabling real-time data, either the ontology is missing the RabbitMQ linking data or no messages are being published. Inspect a dump file to verify the data.

3D Visualization (.pck resource pack)

1. Create a separate Godot project (same Godot version as DTInsight).
2. Create a scene called main.tscn with a _on_message(message) function to receive real-time data.
3. Export the project as a .pck file.
4. Ensure the OML description includes DTDFVocab:HasVisualization true.
5. In DTInsight, pick the .pck file to open the 3D visualization in a pop-up.

The visualization reflects incoming sensor data (e.g., updated temperature labels, heater color changes).

Viewing Component Scripts

If the ontology contains relative file paths to DT source code:

1. Select the software folder from the settings panel.

   

2. Components with associated scripts will display a button with the script filename. Click to view.

   

Continuous Report Generation (CI/CD)

DTInsight integrates into a CI/CD pipeline (currently GitHub Actions) to automatically produce an up-to-date reporting page on every commit. The pipeline:

1. Loads the committed DTDF ontology into a Fuseki server.
2. Runs DTInsight (headless, via Linux xvfb) and triggers the internal TCP CI/CD server (port 9090).
3. Outputs:
   • HTML characteristics table — summary of the 21 DTDF characteristics
   • Architecture screenshot — high-resolution tiled capture of the DT constellation
   • YAML architecture dump — machine-readable description of the constellation
4. Deploys the reporting page as a static website (via Hugo).

An example reporting page for the incubator DT is available at: oakeslabmtl.github.io/DTDF

Steps to set up continuous reporting for your DT

1. Model your DT in the DTDF using OML/OWL/RDF.
2. Configure the static website deployment (e.g., Hugo + GitHub Pages).
3. Set up the CI/CD workflow to load the ontology, run DTInsight, and deploy.

Related Resources

• DTDF Ontology & Example DT: oakeslabmtl/DTDF
• openCAESAR Rosetta: opencaesar/oml-rosetta
• Example DT (Incubator): INTO-CPS-Association/example_digital-twin_incubator

Citation

If you use DTInsight in your research, please cite our paper. An open-access preprint is available on arXiv.

@INPROCEEDINGS{fiter2025dtinsight,
    author={Fiter, Kérian and Malassigné-Onfroy, Louis and Oakes, Bentley},
    booktitle={2025 ACM/IEEE 28th International Conference on Model Driven Engineering Languages and Systems Companion (MODELS-C)}, 
    title={DTInsight: A Tool for Explicit, Interactive, and Continuous Digital Twin Reporting}, 
    year={2025},
    volume={},
    number={},
    pages={139-143},
    keywords={Visualization;Three-dimensional displays;Systematics;Data visualization;Ontologies;Real-time systems;Software;Digital twins;Stakeholders;Monitoring;digital twins;software visualization;software documentation;decision-making;ontologies;OML;monitoring},
    doi={10.1109/MODELS-C68889.2025.00030}
}

Credits

• Kérian Fiter (@KerianFiter)
• Louis Malassigné-Onfroy (@Ryskann)
• Under the supervision of Prof. Bentley Oakes

License

This project is licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.