GenFacet: End-to-End Generative Faceted Search via Multi-Task Preference Alignment in E-Commerce DOI: XXXXXXX.XXXXXXX Conference: Proceedings of the 49th International ACM SIGIR Conference on Research and Development in Information Retrieval; July 12–16, 2026; Melbourne, VIC, Australia ISBN: 978-1-4503-XXXX-X/18/06 CCS: Information systems Retrieval models and ranking CCS: Applied computing Online shopping CCS: Computing methodologies Natural language processing
Zhouwei Zhai, Min Yang, Jin Li · 2026
Read GenFacet: End-to-End Generative Faceted Search via Multi-Task Preference Alignment in E-Commerce DOI: XXXXXXX.XXXXXXX Conference: Proceedings of the 49th International ACM SIGIR Conference on Research and Development in Information Retrieval; July 12–16, 2026; Melbourne, VIC, Australia ISBN: 978-1-4503-XXXX-X/18/06 CCS: Information systems Retrieval models and ranking CCS: Applied computing Online shopping CCS: Computing methodologies Natural language processing by Zhouwei Zhai, Min Yang, Jin Li (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-19665.txt.
Description
Zhai and co-authors present GenFacet, a generative framework for faceted search in e-commerce, deployed at JD.com. Facets guide shoppers through huge catalogues, but rule-based systems age poorly. GenFacet treats the task as two linked generation jobs in one language model. One job creates trend-aware navigation facets from context. The other rewrites user picks into precise search queries. Training mixes teacher-student distillation with GRPO to match user preferences. Online tests show facet clicks up 42 percent and conversions up 2 percent.
Catalog record
| CREATOR | Zhouwei Zhai, Min Yang, Jin Li |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 21594 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.19665