Recommendation or Ranking Prototype Plan
Plan a simple recommender and its evaluation.
The Open buttons send the prompt text as currently shown below to that site only when you click. Fill in the placeholders first.
Plan a prototype recommender. Items, users, and interactions available: [ITEMS_USERS_INTERACTIONS]. Business goal: [BUSINESS_GOAL]. Constraints (cold start, privacy, catalogue changes): [CONSTRAINTS]. Produce a staged plan: 1. Baselines: most popular items, recently popular items, and "customers who bought X also bought Y" using co-occurrence counts with lift = P(A and B) / (P(A) x P(B)). Give the code. 2. Content-based option using item attributes and similarity. 3. Collaborative filtering with matrix factorization, with the data requirements. 4. Cold-start handling for new users and items. 5. Offline evaluation: leave-last-out split by time, metrics (hit rate at k, NDCG at k, coverage, novelty), and the caveat that offline results may not predict online behavior. 6. Online test design with guardrails (conversion, complaints). 7. Risks: feedback loops, popularity bias, filter bubbles, and use of sensitive attributes [VERIFY in [JURISDICTION]]. 8. A table of decisions to make with defaults. Start with the simplest baseline and say when to stop.
Fill in before use
Optional: type your details and the prompt above updates. Nothing is saved or sent.
Replace each bracketed placeholder with your own details, then paste into Any AI or another assistant.