AWS personalize limit on number of items considered when training

Question

Amazon personalize puts a limit of 750,000 on the number of items considered by models during the training process. I have roughly 2 million items (videos) with no interaction but I want to somehow show these items to users. The first approach I considered was to randomly mix these cold items with the ones coming from the recommendation model. The problem is these cold items might be irrelevant to the user so the user will not interact with the item and it still remains cold. What can I do to make these random recommendations of cold items a bit more relevant to the user? (I prefer to only use AWS services since I can't train a model myself)

Answer 1

The user-personalization (custom), top-picks-for-you (VOD), and recommended-for-you (e-comm) recipes all support automatic exploration of new/cold items. Your model is still limited to 750K items but Personalize will automatically include cold items in recommendations that are still relevant based on item metadata (if present in your dataset group). You control what items are considered cold by the item's creation timestamp and the item age cutoff when creating the campaign/recommender. Personalize performs auto-updates every two hours to bring in new cold items for exploration. Regular full retraining is still recommended. This won't get all 2M items visible in the model but it will get cold items in front of your users in a relevant way.