Research map / Learning theory and algorithms / Prediction and decision losses
Learning to Suggest Breaks: Sustainable Optimization of Long-Term User Engagement
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use learned, individualized Lotka–Volterra break policies instead of no breaks | Recommendation policy · MovieLens 1M | mean LTE: +5.98{}% | Helped |
| Update the learned policy at T₀=5 using additional user ratings | LV · MovieLens 1M | mean LTE: +6.35{}% relative to default; LV is +5.98{}% | Helped |
| Use a preemptive learned break schedule rather than a consumption-threshold safety switch | Recommendation policy · MovieLens 1M | mean LTE: +5.98{}% vs. +5.74{}% | Helped |
| Choose a break rate by optimizing the fitted engagement curve rather than selecting the best observed rate | Recommendation policy · MovieLens 1M | mean LTE: +5.98{}% vs. +2.05{}% | Helped |
| Use threshold τ=14 instead of a preemptively optimized break schedule | Recommendation policy · MovieLens 1M | mean LTE | Hurt |
| Increase the experimental break rate p₁ | LV · MovieLens 1M | mean LTE | Helped |
| Use too small an experimental break rate when prediction accuracy is low | LV · MovieLens 1M | mean LTE | Hurt |
| Adapt the policy with mild feedback (ρ=0.15) at T₀=5 | LV-adaptive · MovieLens 1M | mean LTE | Helped |
| Update with too little feedback (ρ<0.15) at T₀=5 | LV-adaptive · MovieLens 1M | mean LTE | Hurt |
| Update very early at T₀=0.5 without very frequent feedback | LV-adaptive · MovieLens 1M | mean LTE | Hurt |
| Update late at T₀=50 | LV-adaptive · MovieLens 1M | mean LTE | No clear effect |
Closest papers by meaning
- Fighting Boredom in Recommender Systems with Linear Reinforcement Learning (NeurIPS 2018 · similarity 0.88)
- Learning the Optimal Recommendation from Explorative Users (AAAI 2022 · similarity 0.88)
- Recommender Ecosystems: A Mechanism Design Perspective on Holistic Modeling and Optimization (AAAI 2024 · similarity 0.86)
- MultiScale Contextual Bandits for Long Term Objectives (NeurIPS 2025 · similarity 0.86)
- Preference Diffusion for Recommendation (ICLR 2025 · similarity 0.86)
- Influential Exemplar Replay for Incremental Learning in Recommender Systems (AAAI 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.