Knowledge Graph Embedding and Graph-Based Models for Leak Testing System

This repository implements several Knowledge Graph Embedding (KGE) and Graph-Based models to analyze data from a leak testing system. The models are trained and evaluated on a dataset derived from the original Neo4j database of the leak testing system.

---

Repository Structure

• Main.py: The main script to train and evaluate the models.
• dataset/: Contains the training, validation, and test datasets.
  • train_enhanced.txt: Training dataset.
  • valid_enhanced.txt: Validation dataset.
  • test_enhanced.txt: Testing dataset.
• results_1000.json: Results for Knowledge Graph Embedding models.
• results_graph_1000.json: Results for Graph-Based models.

---

Prerequisites

• Python >= 3.8
• Required packages listed in requirements.txt

---

Installation

1. Clone the repository: git clone https://github.com/yourusername/your-repo-name.git cd your-repo-name

2. Install the required Python packages: pip install -r requirements.txt

---

Dataset Description

• Original Dataset: Derived from the leak testing system and the corresponding Neo4j knowledge graph database.
• Enhanced Dataset: Preprocessed and split into:
  • train_enhanced.txt (training)
  • valid_enhanced.txt (validation)
  • test_enhanced.txt (testing)

---

Running the Main Script

Main.py performs the following steps:

1. Loads the dataset from the dataset/ folder.
2. Trains several KGE and Graph-Based models.
3. Evaluates the models on the test dataset.
4. Saves the results to results_1000.json and results_graph_1000.json.

Run the script with: python Main.py

---

Results

• results_1000.json: Contains evaluation metrics (e.g., Hits@K, MRR) for Knowledge Graph Embedding models.
• results_graph_1000.json: Contains evaluation metrics for Graph-Based models.

---

Models Trained

Knowledge Graph Embedding Models:

• TransE
• RotatE
• ComplEx
• DistMult
• TransH
• TransR
• TransD

Graph-Based Models:

• GraphSAGE
• R-GCN
• A2N
• CompGCN
• SE-GNN
• Convolutional GNN (if applicable)

---

Contributing

Feel free to open an issue or submit a pull request if you would like to contribute to the project.

---

License

This project is licensed under the MIT License.