Problem
Research questions and friction points this paper is trying to address.
Personal-Agent Mediated Recommendation
Cross-Platform User History
Recommendation Mediation
Rescue-Harm Trade-off
User-Governed Personalization
Personal-Agent Mediated Recommendation with Cross-Platform User History
🤖 AI Summary
This study addresses the challenge in cross-platform recommendation where personal agents, leveraging authorized histories to mediate rankings, struggle to balance beneficial corrections against harmful overrides. To this end, we propose Personal Attribution-Mediated Optimization (PAMO), a theoretical framework and corresponding algorithm. Built upon large language models (LLMs) and reinforcement learning, PAMO incorporates a counterfactual masking mechanism for cross-platform data fusion and introduces the MediateRec benchmark, enabling LLM-based agents to intelligently mediate platform ranking outcomes while establishing the method's local optimality. Experimental results demonstrate that PAMO significantly outperforms baselines on both seen and unseen platforms, effectively achieving a superior rescue-harm trade-off. Ultimately, this work establishes a novel paradigm for personalized mediated recommendation.
📝 Abstract
Modern recommendation is shifting from platform-centric personalization toward user-governed personalization, where a personal LLM agent can act on the user's behalf across services. We formalize this emerging paradigm as Personal-Agent Mediated Recommendation: a platform recommender ranks a candidate set using platform-local information, and a personal agent uses user-authorized cross-platform history to mediate the resulting ranking and produce the final top-K slate. Such mediation is nontrivial: the platform ranking can encode strong population evidence that the personal agent cannot observe, so effective mediation must therefore balance beneficial rescues against harmful overrides. To study this trade-off, we introduce MediateRec, a benchmark that includes scalable proxy cross-platform environments and a real cross-platform test under a controlled platform-agent information boundary. To train the agent to use cross-platform history effectively, we further propose Personal Attribution Mediation Optimization (PAMO), which counterfactually masks that history to estimate personal mediation support and reallocates rank-aware advantage mass under a platform-relative value floor. We theoretically prove that PAMO preserves cutoff-level advantage mass and is locally optimal among first-order reallocations that preserve this mass without lowering average platform-relative value. Experiments on MediateRec show that personal-agent mediation enables meaningful platform corrections, yet even strong proprietary LLMs introduce non-negligible harmful overrides. PAMO consistently improves over matched outcome-only RL across seen and unseen target platforms and on the real cross-platform test, while achieving a better rescue-harm balance.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Personal-Agent Mediated Recommendation
Cross-Platform User History
MediateRec Benchmark
Personal Attribution Mediation Optimization (PAMO)
Counterfactual Masking
🔎 Similar Papers
No similar papers found.
