🎧 FairPlay — Responsible Music Recommendations

FairPlay is a Responsible AI music recommendation system that helps users understand and correct popularity bias in algorithmic music discovery.

Instead of treating recommendation algorithms as black boxes, FairPlay:

• Audits a user’s listening behavior
• Quantifies how much discovery they’re actually getting
• Generates transparent, fairness-aware recommendations that preserve taste while elevating emerging artists

Every recommendation comes with a clear explanation, and users can directly control how strongly fairness is applied.

---

🌍 Inspiration

Music recommendation systems shape what artists get heard—but often invisibly.
While using Spotify, I noticed that even with diverse tastes across genres like dancehall, afrobeats, and soca, recommendations still leaned heavily toward familiar or already popular artists.

This raised a simple but powerful question:

How much discovery am I actually getting—and who is being left out by default?

FairPlay was built to make recommendation systems more transparent, accountable, and human-centered, especially in creative spaces where exposure matters.

---

🎯 What FairPlay Does

1️⃣ Listening Bias Audit

FairPlay analyzes recent Spotify listening history and computes:

• Total tracks and unique artists
• Genre diversity
• Popularity distribution:
  • Mainstream
  • Mid-popularity
  • Emerging
• An Exploration Score that quantifies discovery behavior

2️⃣ Fairness-Aware Recommendations

Instead of optimizing purely for popularity:

• Candidate tracks are generated from the user’s actual genre preferences
• Artists are re-ranked to counter popularity bias
• Emerging and mid-popularity artists are intentionally elevated

3️⃣ Full Transparency

Each recommendation includes a clear explanation such as:

• New vs familiar artist
• Genre match
• Artist popularity tier

No hidden logic. No black box.

4️⃣ User Control

Users can tune how strongly fairness is applied via a fairness_strength parameter—making the algorithm accountable and adjustable.

---

🧠 How It Works

1. Spotify Authentication

   • Users authenticate via Spotify OAuth
   • Recent listening history is fetched securely

2. Audit Phase

   • Builds a listening profile from recent tracks
   • Computes popularity bias and exploration metrics

3. Candidate Generation

   • Extracts top genres from recent listening
   • Searches Spotify for tracks in those genres
   • Avoids popularity-only ranking

4. Fair Ranking

   • Scores tracks using:
     • Genre relevance
     • Artist novelty
     • Popularity fairness adjustment

Mathematically, fairness is introduced as:

$$ \text{score} = \text{relevance} + \alpha \cdot \left(\frac{100 - \text{artist popularity}}{50}\right) $$

Where \(\alpha\) is a user-controlled fairness parameter.

---

📊 Example Output

{
  "top_genres_used": ["dancehall", "afrobeats", "soca"],
  "analysis": {
    "bias_summary": "Your listening is fairly balanced but leans toward familiar artists (76.0%).",
    "exploration_score": 40.4
  },
  "recommendations": {
    "new_artist_rate": 100.0,
    "avg_recommended_popularity": 28.2,
    "items": [
      {
        "track_name": "The System",
        "artist_name": "Courtney Melody",
        "score": 7.54,
        "reason": "new artist; genre match: dancehall, reggae; emerging (23%)"
      }
    ]
  }
}

---

⚖️ Responsible AI Alignment

FairPlay is intentionally designed around Responsible AI principles:

✅ Transparency

Every recommendation includes human-readable reasoning Candidate generation logic is explicit and explainable

✅ Fairness

Actively counteracts popularity bias Promotes discovery of emerging and underrepresented artists

✅ Accountability

User-adjustable fairness strength No opaque ranking logic

✅ Human-Centered Design

Preserves user taste through genre matching Balances relevance with ethical discovery

---

🛠 Tech Stack

Language: Python 3.12

Framework: FastAPI

APIs: Spotify Web API

Authentication: OAuth 2.0

HTTP Client: httpx

Environment Management: python-dotenv

---

📦 Installation & Setup

Run in your terminal:

1️⃣ Clone the Repository

git clone https://github.com/moby7972/moby7972-colorstackwinterhack2025--Fairplay-.git

2️⃣ Create and Activate Virtual Environment

python -m venv .venv

.\.venv\Scripts\Activate.ps1

3️⃣ Install Dependencies

pip install fastapi uvicorn httpx python-dotenv

4️⃣ Create .env File

Create a .env file in the project root with the following values:

SPOTIFY_CLIENT_ID — from Spotify Developer Dashboard
SPOTIFY_CLIENT_SECRET — from Spotify Developer Dashboard
SPOTIFY_REDIRECT_URI — must match the redirect URI configured in your Spotify app

Example:

SPOTIFY_CLIENT_ID=your_client_id
SPOTIFY_CLIENT_SECRET=your_client_secret
SPOTIFY_REDIRECT_URI=http://127.0.0.1:8000/auth/callback

5️⃣ Run the Server

uvicorn app.main:app --reload

6️⃣ Authenticate with Spotify

Open in your browser:

http://127.0.0.1:8000/auth/login

🔗 API Endpoints

Endpoint	Description
/auth/login	Spotify OAuth login
/analyze_spotify	Audit listening behavior
/auth/recommendation_candidates	Generate fair candidate tracks
/recommend_spotify	Full audit + recommendations
/recommend_spotify?fairness_strength=2.0	Stronger fairness bias

🎥 Demo

Video Demo: https://youtu.be/JL5spFgamjA GitHub Repo: https://github.com/moby7972/moby7972-colorstackwinterhack2025--Fairplay-.git

👤 Author

Built by Moby | ColorStack Winter Hackathon 2025