How to get rid of popular items in the recommendation result of AWS Personalize

Question

We are using AWS Personalize to generate frequently bought together recommendation for our products. The algorithms we are using are "Frequently bought together" from "E-commerce recommenders" and "aws-similar-items" recipe.

For both of them, there are popular items which have been returned together with the real recommendations. If I check the document in "https://docs.aws.amazon.com/personalize/latest/dg/native-recipe-similar-items.html", it says: "You can get recommendations for items that are similar to a cold item (an item with fewer than five interactions). If Amazon Personalize can't find the item ID that you specify in your recommendation request or batch input file, the recipe returns popular items as recommendations."

So, we are now wondering, is there a way to let AWS personlize stop returning the popular items for these two algorithms? We only want to show the real similar or frequently bought together recommendations not the popular ones.

Answer

Hello,

I understand that you would like to know if we can stop personalize solution from returning most popular items during recommendations.

Please note that as of now, personalize model has to return any recommendations for the given input. Hence, if sufficient user behavior data for an item isn't available, or if the specified item ID isn't found, the recipe returns popular items as recommendations.

To make sure popular items are not returned. It is recommended to add sufficient user behavior data and the specific items data so that we get item recommendations instead of general popular items data. Unfortunately, as of now we do not have any option to make sure that the personalize solution does not recommend popular data. It has to return recommendations data or popular items data.

How to get rid of popular items in the recommendation result of AWS Personalize

관련 콘텐츠