- 최신
- 최다 투표
- 가장 많은 댓글
Hi, if you add an "author" or "category" field to your item dataset, you can use either filters (to modify all the recommended items) or promotions (apply a filter to a percentage of the recommended items, for instance "have 20% of the recommendations be of followed users"). You would use a filter with a "$variable" ($author or $category) and you would provide the followed authors or categories when you do the recommendation request. You would have to have a database, for instance on dynamoDB to get for the current user, what are the followed authors and categories ad you currently cannot save this data directly in amazon Personalize.
With this methodology, you would be able to guarantee followed authors or posts are always part of the recommended items. It is the most relevant content for those authors or categories for each user that gets recommended.
This blog explains how to use filters and promotions.