Process Mining with Graph Neural Networks

An advanced implementation combining Graph Neural Networks, Deep Learning, and Process Mining techniques for business process analysis and prediction.

1. Overview

This research project implements a novel approach to process mining using Graph Neural Networks (GNN) and deep learning techniques. The framework combines state-of-the-art machine learning models with traditional process mining methods to provide comprehensive process analysis and prediction capabilities.

2. Authors

• Somesh Misra @mathprobro
• Shashank Dixit @sphinx
• Research Group: ERP.AI Research

3. Key Components

1. Process Analysis

• Advanced bottleneck detection using temporal analysis
• Conformance checking with inductive mining
• Cycle time analysis and prediction
• Transition pattern discovery
• Spectral clustering for process segmentation

2. Machine Learning Models

• Graph Attention Networks (GAT) for structural learning
• LSTM networks for temporal dependencies
• Reinforcement Learning for process optimization
• Custom neural architectures for process prediction

3. Visualization Suite

• Interactive process flow visualization
• Temporal pattern analysis
• Performance bottleneck identification
• Resource utilization patterns
• Custom process metrics

4. Technical Architecture

src/
├── input/                # input files
├── models/
│   ├── gat_model.py      # Graph Attention Network implementation
│   └── lstm_model.py     # LSTM sequence model
├── modules/
│   ├── data_preprocessing.py  # Data handling and feature engineering
│   ├── process_mining.py     # Core process mining functions
│   └── rl_optimization.py    # Reinforcement learning components
├── visualization/
│   └── process_viz.py        # Visualization toolkit
└── main.py                   # Main execution script

5. Technical Requirements

• Python 3.8+
• PyTorch 1.9+
• PyTorch Geometric
• PM4Py
• NetworkX
• Additional dependencies in requirements.txt

6. Installation

1. Clone the repository:

git clone https://github.com/ERPdotAI/GNN.git
cd GNN

2. Install dependencies:

pip install -r requirements.txt

7. Data Requirements

The system expects process event logs in CSV format with the following structure:

• case_id: Process instance identifier
• task_name: Activity name
• timestamp: Activity timestamp
• resource: Resource identifier
• amount: Numerical attribute (if applicable)

8. Usage

python main.py <input-file-path>

Results are stored in timestamped directories under results/ with the following structure:

results/run_timestamp/
├── models/          # Trained model weights
├── visualizations/  # Generated visualizations
├── metrics/         # Performance metrics
├── analysis/        # Detailed analysis results
└── policies/        # Learned optimization policies

9. Technical Details

Graph Neural Network Architecture

• Multi-head attention mechanisms
• Dynamic graph construction
• Adaptive feature learning
• Custom loss functions for process-specific metrics

LSTM Implementation

• Bidirectional sequence modeling
• Variable-length sequence handling
• Custom embedding layer for process activities

Process Mining Components

• Inductive miner implementation
• Token-based replay
• Custom conformance checking metrics
• Advanced bottleneck detection algorithms

Reinforcement Learning

• Custom environment for process optimization
• State-action space modeling
• Policy gradient methods
• Resource allocation optimization

10. Contributing

We welcome contributions from the research community. Please follow these steps:

1. Fork the repository
2. Create a feature branch
3. Implement your changes
4. Submit a pull request with detailed documentation

11. Citation

If you use this code in your research, please cite:

@software{GNN_ProcessMining,
  author = {Shashank Dixit/Somesh Misra},
  title = {Process Mining with Graph Neural Networks},
  year = {2025},
  publisher = {ERP.AI},
  url = {https://github.com/ERPdotAI/GNN}
}