paperAugust 2026Unreviewed

MetaStrategy: Generative Ranking with Executable LLM Strategies

Chengyu Lai, Jiuning Lin, Zhibo Xiao, Xiaodong Zhu, Ruiquan Lan, Bin Zhang, Zi-Hong Huang, Wendong Zhang, Chuxin Chen, Yinjiang Cai, Shuaihao Zhong, Lingqin Zhang, Di-Min Wang, Jialin Zhu, Hanqian Zhu

View resource

Abstract

Industrial recommender systems rank heterogeneous content under coupled user, business, commercial, and experience objectives. Existing generative ranking methods typically construct item sequences directly, making them difficult to integrate with mature predictive models, operational rules, and field-level guardrails. We present MetaStrategy, a framework that instead generates a structured, executable ranking strategy. Conditioned on request context, a large language model (LLM) policy emits a

Cite

@misc{lai2026metastrategy,
  title = {{MetaStrategy: Generative Ranking with Executable LLM Strategies}},
  author = {Chengyu Lai and Jiuning Lin and Zhibo Xiao and Xiaodong Zhu and Ruiquan Lan and Bin Zhang and Zi-Hong Huang and Wendong Zhang and Chuxin Chen and Yinjiang Cai and Shuaihao Zhong and Lingqin Zhang and Di-Min Wang and Jialin Zhu and Hanqian Zhu},
  year = {2026},
  month = aug,
  eprint = {2608.09440},
  archivePrefix = {arXiv},
  url = {https://www.semanticscholar.org/paper/eefb388ef446ab525230570a9be8cc931e0c2b34}
}