BAIT: Bias Analytics & Investigation Tool

VAST Challenge 2025 - Mini-Challenge 2 Submission

This repository contains the submission for the VAST Challenge 2025, Mini-Challenge 2. The project is a web-based visual analytics tool designed to help a journalist investigate accusations of bias within a government oversight board.

🌐 Live Demo

A live demo of the application is running at: https://bait.ava25.dbvis.de/

📖 About the Project

The Islands of Oceanus are experiencing economic shifts, with a growing tourism industry challenging the traditional fishing-based economy. The Commission on Overseeing the Economic Future of Oceanus (COOTEFOO) is tasked with monitoring these changes. However, it faces accusations from two opposing groups:

• Fishing is Living and Heritage (FILAH): Accuses COOTEFOO of favoring tourism and neglecting the fishing industry.
• Tourism Raises OceanUs Together (TROUT): Accuses COOTEFOO of being biased towards the entrenched fishing industry.

A journalist, Edwina Darling Moray, is investigating these claims and has acquired datasets from both groups, as well as a more comprehensive dataset of her own. This project provides her with a visual analytics tool to explore these datasets, identify biases, and uncover the truth behind the accusations.

✨ Features

• Bias Scale: Visualize committee bias with a tipping scale metaphor based on sentiment analysis.
• Comparative Analysis: Compare FILAH, TROUT, and journalist datasets side-by-side to reveal sampling bias.
• Dynamic Industry Pairing: Automatically identify and visualize the most polarized industries using cosine similarity.
• Individual Activity Analysis: Use a Parallel Coordinate Plot to track and compare member activities across datasets.
• Network Exploration: Investigate entity relationships with an interactive Ego Network graph.
• Geospatial & Temporal Views: Analyze travel patterns through an integrated timeline and map.

💻 Technical Implementation

The project is a full-stack web application composed of a Python backend and a Vue.js frontend.

Backend

• Data Processing: A Python script (load_data.py) cleans and repairs the JSON datasets.
• Database: Data is stored in a Neo4j graph database for efficient relationship querying.
• API: A FastAPI server provides data to the frontend.
• Containerization: The backend is containerized using Docker for easy deployment.

Frontend

• Framework: A single-page application built with Vue 3 and TypeScript.
• State Management: Pinia is used for centralized state management.
• Visualizations: D3.js is used for interactive visualizations.
• Styling: Tailwind CSS is used for styling.
• Build Tool: Vite provides a fast development experience.

📊 Data Model

The data is represented as a knowledge graph. The full data model is described in the official VAST Challenge documentation.

🚀 Getting Started

Prerequisites

• Docker and Docker Compose
• (Node.js and npm)

Installation

1. Clone the repo

   git clone https://gitlab.dbvis.de/ava2025/group-2.git

2. Start Containers
   • Navigate to the vast-challenge directory.
   • Start the containers:

     docker-compose up --build

     or alternatively

     docker compose up --build

Troubleshooting

In case the frontend is unable to load correctly:

• Navigate to the vast-challenge/frontend directory.
• Install dependencies:

  npm install

• rebuild the frontend service & start again:

  docker compose build --no-cache frontend
  docker compose up

The application will be available at http://localhost:5173.

🔑 Accessing the Services

• Frontend Application: http://localhost:5173
• Backend API Docs: http://localhost:8080/docs
• Backend Health Check: http://localhost:8080/health
• Neo4j Browser: http://localhost:7474/browser/

Neo4j Database Credentials

• User: neo4j
• Password: ava25-DB!! (as defined in docker-compose.yml)

👥 Authors

• Paul Müller: paul.mueller@uni-konstanz.de
• Tilio Schulze: tilio.schulze@uni-konstanz.de