Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn about a novel queueing model for optimizing content review scheduling in social media platforms through this 49-minute conference talk. Explore how major social media companies handle the massive challenge of reviewing billions of posts daily using AI-human pipelines to identify and remove illegal content such as exploitative or copyrighted material. Discover the key operational challenge where the cost of delaying human review depends on uncertain post view counts that evolve over time, a scenario not addressed by existing queueing literature. Examine the development of a new theoretical framework that captures time-evolving uncertainty in post popularity and understand how an asymptotically optimal scheduling algorithm was created to minimize the social harm caused by delayed content moderation. Analyze real-data simulations demonstrating how this approach consistently outperforms current industry heuristics, providing practical insights for improving content moderation systems at scale.
