Measuring and mitigating fake news on Facebook

Scenario: Policy teams need an overnight view of fake‑news prevalence with very few human reviewers. At the same time, they want a long‑term measurement program and model improvements. You must design a rapid assessment, extrapolate platform‑level prevalence, and lay out an iterative roadmap for detection models. ​ Question 1: With limited reviewers, how would you measure fake‑news impact within a single day? (Hint: ML pre‑labels plus targeted human sampling) Question 2: A 1 000‑post sample shows 10 % fake news. How would you extrapolate and report the overall prevalence? (Hint: confidence intervals, weighted projection) Question 3: Given ample resources, design a robust approach to quantify fake‑news prevalence. (Hint: stratified sampling, user exposure, propagation paths) Question 4: Your detection model misses fake content—how would you iterate? (Hint: hard‑negative mining, active learning, ensemble models)