In modern recommender systems, list-wise reranking serves as a critical phase within the multi-stage pipeline, finalizing the exposed item sequence and directly impacting user satisfaction by modeling complex intra-list item dependencies. Existing methods typically formulate this task as selecting indices from the local input list. However, this approach suffers from a semantically inconsistent action space: the same output neuron (logits) represents different items across different samples, preventing the model from establishing a stable, intrinsic understanding of the items. To address this, we propose GloRank (Global Action Space Ranker), a generative framework that shifts reranking from selecting local indices to generating global identifiers. Specifically, we represent items as sequences of discrete tokens and reformulate reranking as a token generation task. This design effectively decouples the scoring mechanism from the variable input order, ensuring that items are evaluated against a consistent global standard. We further enhance this with a two-stage optimization pipeline: a supervised pre-training phase to initialize the model with high-quality demonstrations, followed by a reinforcement learning-based post-training phase to directly maximize list-wise utility. Extensive experiments on two public benchmarks and a large-scale industrial dataset, coupled with online A/B tests, demonstrate that GloRank consistently outperforms state-of-the-art baselines and achieves superior robustness in cold-start scenarios.

翻译：在现代推荐系统中，列表级重排序作为多阶段流水线的关键环节，通过对复杂列表内项目依赖关系的建模来最终确定曝光项目序列并直接影响用户满意度。现有方法通常将该任务建模为从局部输入列表中选择索引。然而，这种策略存在语义不一致的动作空间问题：相同的输出神经元（logits）在不同样本中代表不同项目，导致模型无法建立稳定且本质性的项目理解。为解决这一问题，我们提出GloRank（全球动作空间排序器），这是一种将重排序从选择局部索引转变为生成全局标识符的生成式框架。具体而言，我们将项目表示为离散标记序列，并将重排序重构为标记生成任务。该设计有效解耦了评分机制与可变输入顺序，确保项目能够基于一致的全局标准进行评估。我们进一步通过两阶段优化流水线增强该框架：监督预训练阶段用于用高质量演示初始化模型，随后基于强化学习的后训练阶段直接最大化列表级效用。在两个公开基准数据集和大型工业数据集上的广泛实验，结合在线A/B测试，表明GloRank在冷启动场景中始终优于最先进的基线方法，并展现出卓越的鲁棒性。