https://github.com/franciscomartinez45/Social-Network-Analysis

Applied to analyze how misinformation propagates within communities. With the goal of addressing health disparities and improving health literacy particularly in minority populations, the project explores both supervised and unsupervised learning approaches to understand patterns in graph-structured data using a custom Graph Attention Network
https://github.com/franciscomartinez45/Social-Network-Analysis

matplotlib ml networkx numpy pytorch

Last synced: 10 months ago
JSON representation

Applied to analyze how misinformation propagates within communities. With the goal of addressing health disparities and improving health literacy particularly in minority populations, the project explores both supervised and unsupervised learning approaches to understand patterns in graph-structured data using a custom Graph Attention Network

• Host: GitHub
• URL: https://github.com/franciscomartinez45/Social-Network-Analysis
• Owner: franciscomartinez45
• Created: 2024-06-04T03:18:07.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2025-03-28T03:45:08.000Z (11 months ago)
• Last Synced: 2025-04-11T00:58:01.439Z (10 months ago)
• Topics: matplotlib, ml, networkx, numpy, pytorch
• Language: Python
• Homepage:
• Size: 207 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# Graph Attention Network 🚀

Supervised Machine Learning sandbox used to explore different message passing and influence algorithms from various synthetic datasets. Used as a means to research Addressing Health Disparities through Improved Health Literacy in Minority Populations using AI/ML Models and Social Network Analysis. Created to analyze and potentially improve how minority populations access and comprehend health infromation by leveraging predictive and pattern detection using Machine Learning.

## Table of Contents

- [Features](#features)

- [Screenshots](#screenshots)

- [Installation](#installation)

- [Usage](#usage)

- [Contributing](#contributing)

- [License](#license)

- [Acknowledgments](#acknowledgments)

## Features

- **Interactive Visualization**: Leverages NetworkX and Matplotlib to visualize graph structures, node influences, and model predictions.

- **SHAP Explainer**: Utilized to measure of feature importance by computing Shapley values

- **Health Equity Focus**: Tailored to study and address health disparities in underserved or misrepresented communities.

- **Misinformation Modeling**: Analyzes and models the spread of health-related information (or misinformation) across social networks, providing actionable insights for intervention strategies.

## Screenshots

1. Sample result of 100 nodes using NetworkX synthetic graph generator

   ![App Screenshot](Sample/graph.jpg)

2. Metrics extracted from running the model

   [Click here to view metrics](Sample/metrics.txt)

3. Training Loss

   ![App Screenshot](Sample/training_loss.jpg)

4. SHAP Feature importance summary

   ![App Screenshot](Sample/shap_summary.png)

5. Node activation

   ![App Screenshot](Sample/node_activation.jpg)

## Installation

To get a local copy up and running, follow these steps:

1. Clone the repository:

   ```bash

   git clone https://github.com/franciscomartinez45/Social-Network-Analysis.git

   ```

## Acknowledgments

- **GTgraph**: This project utilizes [GTgraph](https://github.com/Bader-Research/GTgraph), a suite of synthetic random graph generators developed for the 9th DIMACS Shortest Paths Challenge. GTgraph supports various classes of graphs, including:

  - Input graph instances used in the DARPA HPCS SSCA#2 graph theory benchmark (version 1.0).

  - Erdős-Rényi random graphs.

  - Small-world graphs based on the Recursive Matrix (R-MAT) model.

- **NetworkX**: Utilize synthetic graph generators from NetworkX such as Barabasi-Albert and Zachary's Karate Club