Hello,
sorry to barge in via github, but I was not sure how to contact you otherwise. I got few questions regarding the recommender, I was thinking about using it in a school project.
I was wondering if the recommender got somehow integrated and adopted into the Wikidata. I found out a bachelor thesis regarding A-B testing in production , but I was not able to find, whether it was actually kept in production.
On the other note, I wanted to ask about memory consuption of the creation and running process. In the paper you wrote that about 60 000 000 items result in 1.3 GB memory consuption of the existing tree. Is that also regarding the preprocessing and tree creation? And what would be your memory expectation for the current data size (roughly 105 000 000 items). I am asking if I am able to run the recommender (preprocessing included) on my laptop with 24GB RAM.
Thank you for any feedback possible.