Discovery Algorithm Comparator

A comprehensive JavaFX application for process mining, enabling visualization, analysis, and discovery of process models from Petri nets and event logs with advanced algorithm support and Murata reduction techniques.

---

Features

• Model Visualization: Render Petri nets as interactive graphs
• Log Generation: Generate event logs from Petri net models
• Process Discovery: Mine process models from event logs using Heuristics Miner, Inductive Miner, Evolutionary Tree Miner, and Split Miner
• Comparison: Compare original and discovered models side by side
• Murata Reduction: Advanced silent transition removal using Murata reduction techniques for cleaner model visualization
• Statistics: Detailed before/after comparison of model complexity and reduction effectiveness
• Parameter Customization: Fine-tune mining and log generation with algorithm-specific parameter settings
• Comprehensive File Support: Import/export standard process mining files (.pnml, .xes)

---

Project Structure

src
└── main
    ├── java
    │   └── main
    │       ├── MiningApp.java                       # Main application entry point
    │       ├── PnmlToModelConverter.java            # Converts PNML files to Petri net models
    │       │
    │       ├── algorithms                           # Process mining algorithm implementations
    │       │   ├── MiningAlgorithm.java             # Abstract base class for all algorithms
    │       │   ├── MiningAlgorithmSelector.java     # Factory for instantiating algorithms
    │       │   ├── EvolutionaryTreeMinerAlgorithm.java
    │       │   ├── HeuristicMinerAlgorithm.java
    │       │   ├── InductiveMinerAlgorithm.java
    │       │   └── SplitMinerAlgorithm.java
    │       │
    │       ├── controller                           # JavaFX controllers for UI logic
    │       │   ├── LogExplorerController.java       # Manages the log explorer view
    │       │   ├── MiningController.java            # Main window controller
    │       │   ├── ParametersSettingsController.java # Handles generic parameter settings
    │       │   └── StatisticsController.java        # Manages algorithm comparison and statistics
    │       │
    │       ├── controller/settings                  # Controllers for algorithm-specific settings panels
    │       │   ├── ETMParameterStorage.java         # Data storage for Evolutionary Tree Miner
    │       │   ├── EvolutionaryTreeMinerSettingsController.java
    │       │   ├── HeuristicMinerSettingsController.java
    │       │   ├── InductiveMinerSettingsController.java
    │       │   └── SplitMinerSettingsController.java
    │       │
    │       ├── utils                                # Utility and helper classes
    │       │   ├── ExportController.java            # Handles exporting results (XES, PNML, TXT)
    │       │   ├── FileHandler.java                 # Manages file input/output operations
    │       │   ├── MurataReduction.java             # Implements silent transition removal
    │       │   ├── Utils.java                       # Common utility methods and helpers
    │       │   └── VisualisationController.java     # Controls model visualization
    │       │
    │       └── visualisation
    │           └── PetriNetVisualiser.java          # Handles the rendering of Petri nets
    │
    └── resources
        ├── css                                      # Stylesheets for application styling
        │   ├── log-explorer.css
        │   ├── main.css
        │   ├── parameters-settings.css
        │   └── statistics.css
        │
        ├── view                                     # FXML files defining the user interface
        │   ├── logExplorer.fxml                     # Layout for the log explorer
        │   ├── main.fxml                            # Main application window layout
        │   ├── parameters-settings.fxml             # Layout for the settings dialog
        │   └── statistics.fxml                      # Layout for the statistics comparison window
        │
        └── view/settings                            # FXML files for algorithm-specific parameters
            ├── evolutionary-tree-miner-settings.fxml
            ├── heuristic-miner-settings.fxml
            ├── inductive-miner-settings.fxml
            └── split-miner-settings.fxml

---

User Interface Overview

The application features a modern, intuitive interface with three main views:

Main Application View (main.fxml)

• Dual Model Visualization: Side-by-side display of original and discovered models
• Interactive Controls: Real-time parameter adjustment for log generation and mining
• Algorithm Selection: Dropdown menu with all supported mining algorithms
• Murata Reduction Toggle: Checkbox for enabling/disabling silent transition removal
• Export Options: Multiple format export capabilities (XES, PNML, TXT)
• Status Indicators

Log Explorer View (logExplorer.fxml)

• Interactive list of all traces in the event log
• DComprehensive display of individual trace contents
• Save trace information in readable text format

Algorithm Settings View (parameters-settings.fxml)

• Extensive parameter customization with sliders and checkboxes
• Live value display for all adjustable parameters
• Save and reset to default configurations

---

Statistical Reporting

The Murata reduction provides comprehensive statistics including:

• Structural Comparison: Places, transitions, arcs, and silent transitions before/after reduction
• Reduction Summary: Quantitative analysis of components removed
• Percentage Analysis: Relative reduction effectiveness for each component type
• Final Comparison: Original model vs reduced model structural equivalence

---

Requirements

• Java 11
• Apache IvyDE (for dependency management)
• Grandle (for Split Miner algorithm functionalitiess)
• An IDE with IvyDE plugin (Eclipse) recommended

---

Installation

1. Clone the repository

git clone https://github.com/ivananb/DiscoveryAlgorithmComparator.git
cd DiscoveryAlgorithmComparator

2. Build the application (For running Jar)

If you want to build the JAR yourself from the source code, ensure you have Apache Ant and Ivy installed, then run:

    ant main

or

    ant uber-jar

This will create the executable JAR file in the /target directory.

3. Run the application:

You can run the application in two ways:

• From your IDE: Run the MiningApp.java class as a Java Application
• Using the pre-built JAR file (Command Line): Navigate to the project directory and run the following command

    java -jar target/DiscoveryAlgorithmComparator.jar

---

Usage

1. Loading Models and Logs

• Import PNML File or XES Log: Click "Import File" and select a .pnml or .xes file
• The original Petri net will be visualized in the "Original Model" panel

2. Generating Event Logs

• Set parameters in the control panel (if not selected the system will use dafault values):
  • Number of Traces: Quantity of traces to generate (default: 100)
  • Max Trace Length: Maximum length of each trace (default: 50)
• Click "Generate Event Log" to create synthetic logs using stochastic simulation

3. Process Model Mining

• Select your desired mining algorithm from the dropdown menu
• Configure algorithm-specific parameters using the "Settings" button
• Click "Mine New Model" to discover process models from event logs
• The discovered process model appears in the "Discovered Model" panel

4. Murata Reduction

• Toggle the "Remove Silent Transitions" checkbox to enable Murata reduction
• View the simplified model in the Discovered Model panel

5. Algorithm Performance Comparison

• Click "Show Statistics" after generating a log to open the comparison window.
• Configure metric weights (Fitness, Precision, etc.) to influence the scoring, or keep the default values
• Toggle "Remove Silent Transitions" if you want to simplify models before evaluation
• Click "Run Algorithm Comparison" to execute all algorithms and automatically determine the best one based on a weighted overall score. Results are displayed in a detailed table with the winner highlighted

6. Exporting Results

• Export as XES: Save generated event logs in standard XES format
• Export as PNML: Save discovered process models as PNML files
• Export as TXT: Save log information in plain text format from the Log Explorer

Navigation Controls

• Zoom: Mouse wheel.
• Pan: Middle-click drag or Ctrl + Left-click drag.
• Reset View: Right-click and select "Reset View".

---

Supported Algorithms

• Heuristics Miner
• Inductive Miner
• Evolutionary Tree Miner
• Split Miner

---

Keyboard Shortcuts

• Ctrl + Mouse Wheel: Zoom in/out.
• Middle-click + Drag: Pan.
• Ctrl + Left-click + Drag: Alternative pan.

---

Acknowledgments

• ProM Framework: Process mining infrastructure and algorithms
• Viz.js: Graphviz implementation for JavaScript rendering
• OpenXES: XES standard implementation for event logs
• Murata Reduction Techniques: For advanced Petri net simplification methods
• ProM Reduction Plugin: For the Murata reduction algorithm implementation