Build infrastructure to collect online data (e.g. acceptance / rejection of suggested topics, reports of misclassification) and retrain the ML model