Design a Cold-Start-Aware Recommender

You are the data scientist responsible for the recommender behind a newly launched content app. Traffic is thin — most users have barely interacted with anything, and fresh items are uploaded constantly — so both user-side and item-side cold start are the norm rather than the exception. Design a recommender built from two stages, a candidate generation stage followed by a ranking stage, that keeps working well under those conditions. For each stage, name the specific…