Given a large set of customers and a large set of items, how to make predictions given a model like this one:
As stated in the article:
"Since there are over a million apps in the database, it is intractable to exhaustively score every app for every query within the serving latency requirements (often O(10) milliseconds). Therefore, the first step upon receiving a query is retrieval. The retrieval system returns a short list of items that best match the query using various signals, usually a combination of machine-learned models and human-defined rules"
As it is probably often infeasible to do this brute force, ie score every customer X item combination to get the most "relevant" items for all the customers, what kind of systems are commonly used for this initial step? Or is the brute force approach also feasible in practice when we won't have as strict time requirements as this article propose?