News Recommendation System using Reinforcement Learning

This project implements a personalized news recommendation system using the using multi-armed bandit algorithms. It evaluates and compares algorithms like Linear Upper Confidence Bound, Thompson Sampling, UCB1, and Epsilon-Greedy to model and predict user click behavior. By simulating user interactions, the system compares these algorithms in terms of Click-Through Rate (CTR) and decision efficiency, providing insights into their performance in real-world recommendation scenarios.

Features

• 📊 Multiple Algorithms Compared:
  • LinUCB (contextual bandit with user/item features)
  • Thompson Sampling
  • UCB1
  • Epsilon-Greedy
• User Feedback Loop: Accepts user clicks on recommended articles and updates the model online.
• CTR Tracking: Calculates and displays click-through rate based on real user interactions.
• Performance Visualization via CTR-over-time plots.

Dataset

• The project uses the MIND dataset, which contains millions of news impressions and user click behaviors.
• Due to size and licensing, the dataset is not included in the repository.
• Users should download the dataset from the official MIND dataset site and place it in the designated folder before running the code.

📊 Sample Output

Below is a sample result comparing different Multi-Armed Bandit algorithms on the MIND dataset:

---

🚀 Future Work

• Incorporate time-aware or session-based user modeling for better personalization.
• Extend support for hybrid recommendation (content-based + collaborative).
• Integrate a real-time user interaction loop via web interface.
• Experiment with deeper neural contextual bandits (e.g., Deep LinUCB, DQN).
• Explore fairness and bias mitigation in article selection

---

📚 References

1. Lihong Li, Wei Chu, John Langford, Robert E. Schapire.
   A Contextual-Bandit Approach to Personalized News Article Recommendation.
   Proceedings of the 19th International Conference on World Wide Web (WWW), 2010.
   https://arxiv.org/pdf/1003.0146.pdf

2. Lihong Li, Wei Chu, John Langford, Xuanhui Wang.
   Unbiased Offline Evaluation of Contextual-bandit-based News Article Recommendation Algorithms.
   Proceedings of the 16th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD), 2011.
   https://arxiv.org/pdf/1003.5956.pdf