RGCN for Entity Classification

This project implements a Relational Graph Convolutional Network (RGCN) for entity classification using the DGL library. The model is trained and evaluated on various datasets, including AIFB, MUTAG, BGS, and AM. based on Modeling Relational Data with Graph Convolutional Network

Project Structure

• trainUtility.py: Contains utility functions for loading datasets and preparing graphs for training.
• train.py: Main script for training and evaluating the RGCN model.
• RgcnModel.py: Defines the RGCN model architecture.
• LoadDataset.py: Contains functions to load different datasets.
• Layer.py: Implements the RGCN layer.
• requirements.txt: Lists the required Python packages.

Installation

To install the required packages, run:

pip install -r requirements.txt

Usage

To train and evaluate the model on a specific dataset, run:

python train.py --dataset <dataset_name>

Replace <dataset_name> with one of the following: aifb, mutag, bgs, am.

To plot the best validation accuracies for all datasets, run:

python train.py --plot --save_plot <file_name>

Replace <file_name> with the desired file name for the plot image.

Plot

Below is an example plot of the best validation accuracies for all datasets:

License

This project is licensed under the MIT License.