Point-of-Interest Recommendation via Graph Neural Networks

A Master's thesis project focusing on developing a recommender system using Graph Neural Networks on heterogeneous graphs.

The pipeline illustrating the process of recommending points of interest using Graph Neural Networks on heterogeneous graphs.

Project Overview

This project presents the outcomes of a Master’s thesis and internship conducted at Universidad Autónoma de Madrid. The primary objective was to develop a recommender system utilizing Graph Neural Networks (GNNs) applied to heterogeneous graphs, specifically targeting Point-of-Interest (POI) recommendations.

Features

  • Graph Neural Network Implementation: Developed models employing GNNs to capture complex relationships within heterogeneous graph structures.
  • POI Recommendation: Focused on recommending locations of interest to users based on their preferences and behaviors.
  • Data Analysis: Conducted comprehensive analyses on datasets, including the Yelp dataset, to extract meaningful insights.
  • Model Evaluation: Implemented both classification and regression models to evaluate the performance of the recommender system.

Usage

To explore and utilize the project:

  1. Clone the Repository: 
    git clone https://github.com/DarioDiPalma-DDP/Point-Of-Interest_recommendation-through-GNNs-on-heterogeneous-graphs.git
  2. Set Up the Environment: Ensure Python 3.8 or later is installed. Create a virtual environment and install the necessary dependencies: 
    python3 -m venv venv
source venv/bin/activate
pip install --upgrade pip
pip install -r requirements.txt
  3. Explore Notebooks: The repository includes several Jupyter notebooks:
    • Model_V2_Classification.ipynb: Contains the classification model implementation.
    • Model_V2_Regression.ipynb: Contains the regression model implementation.
    • Yelp Data Analysis.ipynb: Provides data analysis on the Yelp dataset.

    Open these notebooks using Jupyter to review the code and results.

Dependencies

  • Python: Version 3.8 or later
  • PyTorch: For building and training neural network models
  • PyTorch Geometric: Extension library for GNNs
  • Pandas: Data manipulation and analysis
  • NumPy: Numerical computing
  • Jupyter Notebook: For interactive code execution and visualization

For detailed information and updates, please refer to the GitHub repository.