1
$\begingroup$

I would like to build a recommendation system based only in the items metadata.

I have an input vector with some desirable topics that the user want to read about, for example: (self-help, yoga, sports)

On the other hand I have a dataset with books with Title, Description, among other fields.

Up to now I am building this model locally with Python, using clustering to group books by similarity.

However, I would like to build a Recommendation System using Amazon Personalize for this case using only item metadata. I do not want to add other info as ratings for now.

Is it possible? Do you know some example?

| improve this question | | | | |
$\endgroup$
0
$\begingroup$

As far as I have researched, Amazon Personalize is not useful for this project.
This is because Amazon Personalize requires User-Item Interaction data to train the model. To use Amazon Personalize, you need a dataset with historical data that contains at least the fields:

User 
Item 
Date 

where the Date field represent the time when the user interacted with the Item.

| improve this answer | | | | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.