Large language models (LLMs) are promising backbones for generative recommender systems, yet a key challenge remains underexplored: verbalization, i.e., converting structured user interaction logs into effective natural language inputs. Existing methods rely on rigid templates that simply concatenate fields, yielding suboptimal representations for recommendation. We propose a data-centric framework that learns verbalization for LLM-based recommendation. Using reinforcement learning, a verbalization agent transforms raw interaction histories into optimized textual contexts, with recommendation accuracy as the training signal. This agent learns to filter noise, incorporate relevant metadata, and reorganize information to improve downstream predictions. Experiments on a large-scale industrial streaming dataset from Netflix show that learned verbalization delivers up to 93% relative improvement in discovery item recommendation accuracy over template-based baselines. Further analysis reveals emergent strategies such as user interest summarization, noise removal, and syntax normalization, offering insights into effective context construction for LLM-based recommender systems.

翻译：大语言模型（LLMs）是生成式推荐系统的有前途的基础架构，但一个关键挑战仍未被充分探索：语言化，即将结构化用户交互日志转化为高效的自然语言输入。现有方法依赖仅串联字段的固定模板，生成的表示对推荐而言非最优。我们提出一个数据为中心的框架，用于学习基于LLM的推荐系统语言化。通过强化学习，语言化智能体将原始交互历史转化为优化的文本上下文，以推荐准确性作为训练信号。该智能体学会过滤噪声、整合相关元数据并重组信息以改进下游预测。在Netflix的大规模工业流式数据集上的实验表明，与基于模板的基线相比，学习的语言化在发现项目推荐准确性上实现了高达93%的相对提升。进一步分析揭示了如用户兴趣总结、噪声移除和语法规范化等涌现策略，为基于LLM的推荐系统有效上下文构建提供了洞见。