News Article Recommendation System

---

Old Version — Streamlit + Heroku

Built with: Python, NumPy, Pandas, Seaborn, Matplotlib, Plotly, NLTK, scikit-learn, Pickle, Streamlit Tools: Jupyter Notebook, Spyder, Docker, Heroku Website: news-article-recommender.herokuapp.com

Goal: Recommend articles that best fit user preferences.

Highlights:

1. Developed a web scraper using Cypress to retrieve story details from The New Yorker website.
2. Fetches 90 news articles daily and appends new articles every other day.
3. Automated web scraping with a CI/CD pipeline.
4. Made pipeline OS-independent using Cypress Docker containerization.
5. Employed TF-IDF and Word2Vec embeddings for article recommendations.
6. Found Word2Vec performed better in capturing semantic similarity.
7. Deployed the application on Heroku and Docker for ease of access.

---

New Version — FastAPI + HTML + Google Cloud Run

A modern web application that recommends news articles based on a headline input using FastAPI and TF-IDF similarity, with a clean HTML/CSS/JavaScript frontend.

---

Features

• Enter a news headline and get real-time article recommendations.
• Backend powered by FastAPI with efficient similarity computations using scikit-learn.
• Minimalistic frontend using plain HTML, CSS, and JavaScript for a responsive user experience.
• Fully containerized with Docker for portability and ease of deployment.
• Easily deployable locally or on cloud platforms like Google Cloud Run.

---

Project Structure

├── Backend/           # FastAPI backend API code
├── Frontend/          # HTML, CSS, and JavaScript frontend files
├── Pickle/            # Serialized data and TF-IDF model pickle files
├── Dockerfile         # Docker configuration for containerizing the app
├── README.md          # Project documentation (this file)
└── requirements.txt   # Python dependencies (optional)

---

Getting Started

Prerequisites

• Python 3.7 or higher
• Docker (for containerized deployment)
• (Optional) Google Cloud SDK for deployment to Cloud Run

---

Run Locally Without Docker

1. Install required Python packages:

   pip install fastapi uvicorn scikit-learn pandas numpy python-multipart

2. Start the FastAPI backend server:

   uvicorn Backend.app:app --reload --port 8080

3. Open Frontend/index.html in your browser to access the UI.

---

Run With Docker

1. Build the Docker image:

   docker build -t news-recommendation-app .

2. Run the container locally:

   docker run -p 8080:8080 news-recommendation-app

3. Open http://localhost:8080 in your browser.

---

Deploy on Google Cloud Run

1. Configure your GCP project and build the Docker image:

   gcloud config set project YOUR_PROJECT_ID
   docker build -t gcr.io/YOUR_PROJECT_ID/news-recommendation-app .
   docker push gcr.io/YOUR_PROJECT_ID/news-recommendation-app

2. Deploy the container to Cloud Run:

   gcloud run deploy news-recommendation-app \
       --image gcr.io/YOUR_PROJECT_ID/news-recommendation-app \
       --platform managed \
       --region YOUR_REGION \
       --allow-unauthenticated \
       --port 8080

3. Access your deployed app via the URL provided by Cloud Run.

---

Architecture

---

Tech Stack

• FastAPI — Fast and modern web API framework
• scikit-learn / nltk — TF-IDF vectorization and cosine similarity calculations
• HTML / CSS / JavaScript — Frontend user interface
• Docker — Containerization for consistent environments
• Google Cloud Run — Serverless container hosting (optional cloud deployment)

---