I have worked on online recommender systems. (Here, “online” means the model is being continuously updated from user interactions.) Overfitting is typically no

I have worked on online recommender systems. (Here, “online” means the model is being continuously updated from user interactions.) Overfitting is typically not a concern. You train on the last N days of user interactions, and because of the volume of data there isn’t time for the model to see an interaction twice. So you don’t need a test set. Your performance metrics may go up or down in a day depending on data drift.