EchoChamber Studio

Simulation of discursive bubbles using political comments. Each agent responds from the perspective of its own political community.

Project Structure

echochamber/
├── notebooks/              # Weekly course notebooks (added during the semester)
├── collector/              # Scripts for collecting comments from YouTube / RSS
├── data/
│   ├── raw/                # Raw collected comments (CSV or JSONL)
│   ├── cleaned/            # Cleaned and standardized corpus
│   └── bubbles/            # One JSONL file per agent after annotation
├── assets/
│   └── roles/              # Agent role cards (roles.yaml) — written by students
├── scripts/
│   ├── clean_corpus.py     # Cleans and standardizes raw data
│   └── build_vectorstore.py # Builds FAISS vector index from data/bubbles/
├── core/                   # Core infrastructure — do not modify
│   ├── agent.py            # Agent class: reads roles.yaml + retrieves from corpus
│   ├── retriever.py        # Semantic search over FAISS index
│   ├── graph.py            # LangGraph agentic debate orchestration
│   └── metrics.py          # Dissimilarity, sentiment, and visualization
├── app/
│   └── app.py              # Gradio application (built incrementally during course)
└── reports/                # Final report and ethics checklist templates

Setup

git clone <your-repo-url>
cd echochamber
python -m venv .venv
source .venv/bin/activate   # Windows: .venv\Scripts\activate
pip install -r requirements.txt
cp .env.example .env        # then add your API key

Team

• Team name:
• Topic / bubble theme:
• Members and agents:
  • Member 1 → Agent:
  • Member 2 → Agent:
  • Member 3 → Agent:
  • Member 4 → Agent:
  • Member 5 → Agent: