[Machine Learning Recommendations API] Recommendations adjustment in realtime

Question

All,

I'm wondering , does Recommendations API can adjust recommendation basing on realtime usage event acquisition from website for instance.

It appears  that API only works with manually uploaded files.

I have sent a lot of usage events (AddUsageEvent) for the different users and Items and according usage statistics (GetUsageStatistics) all those events has been captured by model.

But it didn't change recommendation results somehow, even after model rebuild.

Could someone advice?

Answer 1

The events sent through AddUsageEvents are consumed by the recommendations API. So, assuming the pattern of consumption for your items changed, they should change the model after you rebuild it.

Could you send the model id / build id information to [e-mail]. Please refer to this forum question, and we can dig deeper into why you are not seeing changes...

Thanks!

Luis Cabrera
Program Manager
Recommendations API

---

Luis Cabrera. Program Manager -- Azure Machine Learning @luisito_cabrera Disclaimer: This post is provided AS IS - with no warranties, and confer no rights.

Answer 2

Thanks Luis.

I have sent  an email with details