Extensibility for new techniques
Adding a new discovery algorithm, conformance technique, or visualization to Promenade is a plugin, not a fork.
Process-mining research produces new discovery algorithms, conformance-checking techniques, and visualizations faster than any single tool can ship them by default. Promenade’s response is to make the plugin interface the same interface the built-in algorithms use — there’s no separate, more limited “external plugin API.”
A plugin can:
- Contribute a discovery or conformance action that takes one or more artifacts as input and produces a new artifact, using the WebAssembly, Python (via Pyodide), or relational (SQL-over-DuckDB) runtime, whichever fits the technique.
- Contribute a view for rendering an artifact type, including custom object-centric or object-centric-adjacent visualizations that don’t exist in the core application.
- Read event data and object-centric data through the same host data contract the built-in algorithms use, so a novel technique is not second-class in terms of what data it can see.
This matters specifically for reproducible research: implementing a technique from a paper as a Promenade plugin gives readers of that paper a way to run it themselves in a browser, against their own data, without reconstructing your experimental setup.
See Plugin development for the technical detail of writing one.