Continuous Monitoring for Machine Learning Recommendation?

The monitoring and observability space is really mostly focused on classification and regression. We're still very much developing the best practices around recommendation systems. But there are some fundamental similarities and some best practices that we can very trivially apply from classification to or classification slash regression to recommendation. The more insidious thing that happens is when you don't have like a global pandemic, but very slow societal changes that end up impacting people's preferences. When those types of insidious changes happen, you need to be able to monitor in order to understand those changes.