⚽ GOALIE – Game-Observations-Analysis-via-Linguistic-Information-Extraction

Real-Time Match Insights from Commentators Narration – All in Your Browser.

GOALIE is a browser extension (designed for Chromium) that simulates real-time football match updates, complete with AI-powered snippet generation, tactical commentary classification, and interactive match displays. It features a sleek UI and integrates with a Flask-based backend that talks to a Gradio-hosted classifier.

---

Features

• 🎮 Live Match Simulation with realistic minute-by-minute commentary (For Demostartion Purposes)
• 💬 AI-Generated Snippets describing the current game state
• 🔍 NLP Classification of football commentary using Hugging Face + Gradio
• 🧠 Event Detection for goals, fouls, corners, substitutions, and more
• ⚙️ Configurable Match Data (currently with limited commentary, in production using SportsMonk API (Extension Under Development))
• 🌙 Dark Themed UI with smooth animations and responsive layout

---

📊 System Pipeline Diagram

Tech Stack

Layer	Technology
Frontend	HTML, CSS (custom + variables), JavaScript
UI Libraries	Feather Icons, Google Fonts (Inter)
Backend	Python, Flask, Flask-CORS
AI/NLP	Hugging Face Transformers (via Gradio client)
Hosting	Local Flask server + Gradio space endpoint

---

Model Benchmarking & Evaluation

GOALIE classifies football commentary into 3 categories — CURRENT, PAST, and IRRELEVANT — using supervised NLP techniques.

---

Dataset

• Source: SoccerNet-Echoes (whisper_v3/en)
• Processed version: football_commentary_balanced.csv (balanced to 50,000 samples/class)
• Final shape: 149,981 samples
• Columns: game, text, label

Used only English commentary. Filtered and balanced across the 3 label classes.

---

Top 10 Model Combinations

Rank	Model	Vectorizer	Accuracy	F1 Score	CV Score (±std)	Time (s)
1	Logistic Regression	TF-IDF (Char 2-5)	0.9966	0.9966	0.9959 ± 0.000	135.03
2	Random Forest	TF-IDF (Char 2-5)	0.9873	0.9874	0.9889 ± 0.001	266.37
3	Logistic Regression	Count (1-2gram)	0.9790	0.9790	0.9791 ± 0.001	21.72
4	Logistic Regression	TF-IDF (1-3gram)	0.9789	0.9789	0.9774 ± 0.001	17.25
5	Logistic Regression	TF-IDF (1-2gram)	0.9773	0.9773	0.9757 ± 0.001	15.50
6	Random Forest	TF-IDF (1-2gram)	0.9340	0.9336	0.9351 ± 0.002	22.90
7	Random Forest	TF-IDF (1-3gram)	0.9332	0.9329	0.9322 ± 0.000	25.37
8	Naive Bayes	TF-IDF (Char 2-5)	0.9331	0.9328	0.9341 ± 0.001	108.53
9	Random Forest	Count (1-2gram)	0.9315	0.9312	0.9329 ± 0.001	30.64
10	Naive Bayes	TF-IDF (1-3gram)	0.8689	0.8646	0.8640 ± 0.002	14.73

---

Best Performing Model

Model     : Logistic Regression
Vectorizer: TF-IDF (Character-level, 2–5)
Accuracy  : 0.9966
F1 Score  : 0.9966
CV Score  : 0.9959 ± 0.000

---

Setup Instructions

1. Clone the repository

git clone https://github.com/16kushaal/GOALIE-Live-Sports-Extension.git
cd GOALIE-Live-Sports-Extension

2. Backend Setup (app.py)

⚠️ Ensure you have Python 3.8+ and pip installed.

pip install Flask Flask-Cors python-dotenv gradio_client

Create a .env file in the root directory:

GRADIO_SPACE=<your_gradio_space_url>
HF_TOKEN=<your_huggingface_token>

Then run:

python app.py

Your API will be hosted at http://127.0.0.1:5000/classify_and_log

---

3. Frontend / Chrome Extension Setup

Load Extension

1. Open Chrome → chrome://extensions
2. Enable Developer Mode
3. Click Load unpacked → select the repo folder

Ensure the following file exists and is correct:

manifest.json

Insert Gemini API Key (Frontend)

In popup.js, replace:

const GEMINI_API_KEY = 'ADD_YOUR_GEMINI_API_HERE';

with your actual API key, or handle it through a proxy if sensitive.

---

Project Structure

FOOTBALL-SCORE-EXTENSION/
│
├── .venv/                          # Python virtual environment
├── .env                            # Environment variables (Gradio + HuggingFace tokens)
├── README.md                       # Project documentation
├── app.py                          # Flask backend server (text classification + Gradio client)
├── text.txt                        # Match event commentary samples
│
├── football-live-extension/       # Chrome Extension Source
│   ├── manifest.json              # Chrome Extension manifest (v3)
│   ├── popup.html                 # Main popup UI HTML
│   ├── popup.css                  # Extension styling (themed, responsive)
│   ├── popup.js                   # Frontend logic, AI snippet updates, match selection
│   └── images/                    # Logos/icons (48x48, etc.)
│
├── model/                         # ML-related assets
│   ├── data/                      # Labeled and balanced datasets
│   │   ├── football_commentary_labelled.csv
│   │   └── football_commentary_balanced.csv
│   ├── results/                   # Outputs (model + visualizations)
│   │   └── output.png             # Example plot/metric image
│   |── football_classifier_model.pkl
│   ├── dataset.ipynb             # Preprocessing, balancing, and exploration
│   └── model.ipynb               # Training and evaluation of the classifier

---

⚙️ How It Works

1. User selects a match from the popup.

2. Commentary events simulate live match progression every few seconds.

3. Commentary is sent via fetch() to Flask (/classify_and_log) where:

   • It's processed by a Gradio-hosted transformer model
   • The classification (Current, Past, Irrelevant) is logged and printed

4. The popup UI updates the snippet and events visually based on the result.

---

Demo Commentary Types

{
  "type": "commentary", "text": "Dembélé hits the post! That was close!"
}
{
  "type": "score", "scorer": "Mbappé", "team": "home", "newScore": [1, 0]
}
{
  "type": "foul", "text": "Yellow card for Caicedo."
}

---

📋 Todo / Roadmap

• Add real-time data API support (from SportsMonk)
• Improve snippet generation using Gemini API (currently simulated)
• Save commentary logs to a database
• Add user team preferences and theming
• Dockerize backend for production deployment

---

How to run it

1️⃣ Clone the Repository (if not already)

git clone https://github.com/16kushaal/GOALIE-Live-Sports-Extension
cd GOALIE-Live-Sports-Extension

2️⃣ Create and Activate Virtual Environment

3️⃣ Install Required Python Packages

pip install -r requirements.txt

4️⃣ Download spaCy Model

python -m spacy download en_core_web_sm

5️⃣ Configure Environment Variables

Create a .env file in your root directory and add the following:

GRADIO_SPACE=<your_gradio_space_url>
HF_TOKEN=<your_huggingface_token>

6️⃣ Run the Server

python app.py

---

Working Demo Video

---

Author

Made with ⚽ and 💻 by @16kushaal

---

📄 License

Apache-2.0 license. See LICENSE for more information.

---