Which mathematical technique is LEAST commonly used in modern retail product recommendation systems?

In today's data-driven retail environment, mathematical algorithms power the product recommendations you see while shopping online. These systems use various approaches to predict what items might interest you based on your browsing history, purchase patterns, and similarity to other shoppers. This poll tests your knowledge about the mathematical foundations behind modern retail recommendation engines.

This is a live public trivia poll on AIPolls.Net. Vote and see real-time results.

Markov Decision Processes for sequential purchase prediction
Collaborative filtering using matrix factorization
Cosine similarity for content-based filtering
Association rule mining (e.g., market basket analysis)