tvaw-lib 1.0.0

TVAW Nets implementation

TVAW Miner: Typed Petri Nets with Variable Weight Arcs

Authors: Sergeev I.I., Lomazova I. A. Scope: Object-Centric Process Mining

Overview

TVAW Lib - an open-source Python library for the discovery, analysis, and evaluation of Typed Petri Nets with Variable Weight Arcs (TVAWs). The tool extends classical Petri net–based process mining by supporting typed tokens and variable arc weights, enabling richer modeling of object-centric and multi-entity processes.

The framework is currently built on top of the PM4Py ecosystem, with planned native integration in future releases.

Motivation

Traditional process discovery techniques often assume homogeneous tokens and fixed arc semantics, limiting their applicability in object-centric settings. TVAWs generalize these models by:

• Introducing typed tokens to distinguish object classes,
• Allowing variable arc weights to capture flexible consumption/production rules,
• Supporting fine-grained behavioral constraints across interacting entities.

This aligns with recent advances in object-centric process mining .

Features

1. Discovery of TVAW models from event logs
2. Conformance checking (precision and other metrics)
3. Metric computation for model quality evaluation
4. Visualization for visual analysis
5. Integration with PM4Py pipelines

Usage

TVAW Discovery & evaluation

In code:

ocel = pm4py.read_ocel(ocel_path)
net, im, fm = discover_tvaw(ocel)

Then visualize with:

tvaw_gviz = visualize_tvaw(net)
ocpn_visualizer.save(tvaw_gviz, f"{out_path_dir}/tvaw.png")

With CLI:

Create n virtual environments:

# For main discovery
python3 -m venv venv
pip3 install -r requirements.txt

# For analysis
python3 -m venv venv1
pip3 install -r precision/requirements.txt

# For testing
python3 -m venv venv2
pip3 install -r requirements_test.txt

source ./venv2/bin/activate && python3 test.py \
    --logs-dir ./test/jsonocel \
    --results-dir ./results/logistics \
    --compare-python ./venv/bin/python  \
    --precision-python ./venv1/bin/python \
    --compare-script ./produce_comparison.py \
    --precision-script ./precision/calc_precision.py

OCEL Generation

Example:

python3 ./gen_tvaw_logs.py test/gen_configs/cars_min.json test/jsonocel/cars_min.jsonocel

Architecture

The tool is structured into the following components:

1. Discovery Module: Constructs TVAWs from event logs
2. Models layer: Defines the base structures
3. Evaluation Module: Computes precision, and structural metrics (fitness is proven to be = 1)
4. Visualization Module: Provides visualization capabilities

Theoretical Foundations

The theoretical basis for TVAWs are defined in the “Typed Petri Nets with Variable Arc Weights” paper by Irina A. Lomazova, Alexey A. Mitsyuk, and Andrey Rivkin. For the details of the discovery approach, see the docs folder

Roadmap

• Full integration into PM4Py
• Advanced visualization of TVAW models
• Performance optimization for large-scale logs
• Support for real-time process monitoring

Contributing

Contributions are welcome. A PR shall contain documentation, review instructions and tests. Each contribution should have a particular defined and developed scope.

Please submit PRs or open issues for discussion.

License

This project is licensed under the MIT License.

MIT License

Copyright (c) 2026

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction...

Acknowledgments

This work is inspired by foundational research in process mining and Petri nets, particularly in object-centric modeling and evaluation frameworks .