STAR: Staged Training with Aligned Reinforcement Learning and Multi-Faceted Distillation for Interpretable E-commerce Relevance.

@inproceedings{wang-2026-star, author = {Chenxu Wang and Jianzhi Shao and Wenyan Li and Jianzhi Shao and Wenyan Li and Ferhan T\"{u}re and Tao Zhang and Chi Zhang and Ferhan T\"{u}re and Tao Zhang}, editor = {Hannah Bast and Alistair Moffat and Falk Scholer and Marc Najork and Min Zhang and Hannah Bast and Marc Najork and Min Zhang}, title = {{STAR: Staged Training with Aligned Reinforcement Learning and Multi-Faceted Distillation for Interpretable E-commerce Relevance}}, booktitle = {{Proceedings of the 49th International ACM SIGIR Conference on Research and Development in Information Retrieval, SIGIR 2026, MelbourneVICAustralia, July 20-24, 2026}}, pages = {4965--4970}, publisher = {ACM}, year = {2026}, url = {https://doi.org/10.1145/3805712.3808444}, doi = {https://doi.org/10.1145/3805712.3808444} }