I'm investigating if we can use the GCP Retail API for our use case.
We have multiple sets of catalogues and respective user events. However I wouldn't want to mix them and keep them segregated for model training purposes.
These catalogue items and events belong to different customers of ours.
All the data is within a single GCP Project. Would it be possible to make the Retail API segregate them somehow, to make sure when we request recommendation for a particular client, it will be based on the catalogue products and user events from this particular client and not mixing them with other client's data?
Will be using these two APIs for importing the data:
https://cloud.google.com/retail/docs/reference/rest/v2/projects.locations.catalogs.userEvents
https://cloud.google.com/retail/docs/reference/rest/v2/projects.locations.catalogs.branches.products...
Thanks!
I can see the user events have `entity` can we perhaps leverage this to achieve what we need?
The problem is I don't see the same `entity` field as something we can also supply when we are importing the products, to make sure the ML models only learn from products and user events for the same entity.
Help is appreciated!
Working on a similar problem, keeping an eye on this thread
Hello, unfortunately you can only have one retail catalog project per Google Cloud project at this time.
The retail catalog project contains the product data (up to 3 branches) and all user event data. The best method to segment different user data within a single project/catalog right now is Entities - but it still only supports a single currency and does not solve your problem.
If you want separate clients to have different data, they need to be in separate Google Cloud projects. I've outlined some more details and limitations on multi-tenancy here.
Another thing to keep in mind if you're using Recommendations AI, the user events only works on Branch 0 right now, so try to use a single branch when building your solution.