representational-process
Representational process mining
Code structure
pieter/*.py
: Python files for Act2Vec, Trace2Vec, Log2Vec and Model2Vec, as well as support filespieter/data files.7z
: input/output models and logs. Unzip locally but do not commit to GitHub (file size is too large)pieter/.ipynb
: experimental notebookspieter/E-mail with other heatmaps.pdf
: preliminary results when playing with some other Model2Vec ideas
Model2Vec possible approaches:
- Based on graph structure or based on Petri net semantics (Markings)?
- Directional or not?
- Identifiers for places or single "PLACE" token?
- Include places altogether or not?
- Identifiers for activities or single "ACT" token?
- Identifiers for invisible activities?
Preliminary experiments with a structural based approach seem to give good results.