Dynamic Contextual Recommendation Engine Architecture

Architect an advanced recommendation system that dynamically adapts to user context, preferences, and temporal variations. Implement a hybrid recommendation approach combining collaborative filtering, content-based methods, and reinforcement learning techniques. Design a modular system capable of real-time personalization, handling cold-start problems, and providing probabilistic uncertainty estimates for recommendations.