Given a dataset of user interactions, write an algorithm to find the most similar items using vector embeddings.
