This repository provides implementations, data utilities, and experiment scripts for research involving Relational Graph Convolutional Networks (RGCNs) and other graph-based learning models to implement edge weights into RGCN.

📌 Key Features

🧠 Regular RGCN Baseline
Located in baselines/standard_rgcn/, this module includes the standard Relational Graph Convolutional Network implementation: the primary baseline for all experiments.

⚗️ Five Experiment Methods
Found inside experiments/, these are the five distinct methods described in the project report.
Each subfolder contains its own scripts, configurations, and detailed instructions.

🛠 Modular Core System
Shared logic is organized into:

• core/: Model definitions and training logic
• utils/: Data processing, logging, and evaluation tools

📁 Repository Structure

bdrp/

├── baselines/ # Regular RGCN baseline implementation

│ └── standard_rgcn/

├── core/ # Core framework modules

├── data/ # Dataset files or processing helpers

├── docs/

│ └── visualization/

├── experiments/ # Five methods used in the report

├── scripts/ # Miscellaneous scripts for automation

├── utils/ # Utility functions and helpers

├── requirements.txt # Dependencies list

└── .gitignore

🚀 Getting Started

1️⃣ Clone the repository git clone https://github.com/stesilva/bdrp.git cd bdrp

2️⃣ Install dependencies

Create and activate a virtual environment: python3 -m venv venv source venv/bin/activate

Then install required packages: pip install -r requirements.txt

🧪 Running Experiments

📍 Baseline: Regular RGCN
Path:
baselines/standard_rgcn/

Run it using scripts provided in this directory.

📍 Five Custom Methods (from the Report)
Path:
experiments/

Each subfolder represents one experiment setup.
To execute a method, navigate to its directory and follow the included README.

Some methods may require:

• Preparing custom datasets
• Selecting configuration files
• Saving outputs into logs/ or results/