Scheduling with Uncertain Holding Costs and its Application to Content Moderation

2025-05-27

AI Summary

In social media content moderation, the delayed-review cost dynamically evolves along user viewing trajectories and is a priori unknown. Method: We formulate a state-dependent, Markov-evolving queue scheduling model and propose the Opportunity-adjusted Remaining Cost (OaRC) index algorithm—the first to incorporate single-task Markov ski-rental modeling into online scheduling, overcoming the suboptimality of classical cμ and cμ/θ rules. Contribution/Results: We theoretically establish an Õ(L^{1.5}√N) regret bound for OaRC, proving its asymptotic optimality and independence from state-space size. Extensive simulations on real-world advertising delivery and UGC moderation scenarios demonstrate that OaRC significantly outperforms industrial baseline policies, enabling context-aware, real-time scheduling of heterogeneous tasks.

Technology Category

• Planning, Routing, and Scheduling: Scheduling under Uncertainty
• Search and Optimization: Algorithm Configuration
• Machine Learning: Online Learning & Bandits

Application Category

• Responsible Web: Human-perceived consequences of algorithmic deployment on the web
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

In content moderation for social media platforms, the cost of delaying the review of a content is proportional to its view trajectory, which fluctuates and is apriori unknown. Motivated by such uncertain holding costs, we consider a queueing model where job states evolve based on a Markov chain with state-dependent instantaneous holding costs. We demonstrate that in the presence of such uncertain holding costs, the two canonical algorithmic principles, instantaneous-cost ($cmu$-rule) and expected-remaining-cost ($cmu/ heta$-rule), are suboptimal. By viewing each job as a Markovian ski-rental problem, we develop a new index-based algorithm, Opportunity-adjusted Remaining Cost (OaRC), that adjusts to the opportunity of serving jobs in the future when uncertainty partly resolves. We show that the regret of OaRC scales as $ ilde{O}(L^{1.5}sqrt{N})$, where $L$ is the maximum length of a job's holding cost trajectory and $N$ is the system size. This regret bound shows that OaRC achieves asymptotic optimality when the system size $N$ scales to infinity. Moreover, its regret is independent of the state-space size, which is a desirable property when job states contain contextual information. We corroborate our results with an extensive simulation study based on two holding cost patterns (online ads and user-generated content) that arise in content moderation for social media platforms. Our simulations based on synthetic and real datasets demonstrate that OaRC consistently outperforms existing practice, which is based on the two canonical algorithmic principles.

Problem

• Optimizing content moderation scheduling with uncertain holding costs
• Addressing suboptimality of canonical algorithms in dynamic job states
• Developing OaRC algorithm for adaptive future job opportunities

Innovation

• Uses Markov chain for job state evolution
• Develops OaRC index-based algorithm
• Achieves asymptotic optimality with OaRC