Controlling Output Rankings in Generative Engines for LLM-based Search
Haibo Jin, Ruoxi Chen, Peiyan Zhang, Yifeng Luo, Huimin Zeng, Man Luo, Haohan Wang · 2026
Read Controlling Output Rankings in Generative Engines for LLM-based Search by Haibo Jin, Ruoxi Chen, Peiyan Zhang, Yifeng Luo, Huimin Zeng, Man Luo, Haohan Wang (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-2602-03608.txt.
Description
Shoppers increasingly ask a language model what to buy, and the model's shortlist decides who gets seen. Haibo Jin and 3 co-authors present CORE in a 2026 paper, a method for controlling output rankings inside such generative search engines. Because the model's interaction with the search engine is a black box, CORE works on the retrieved content itself.
The method appends crafted optimization text to search results in three styles: plain strings, reasoning passages, and review-like text. To test it, the authors build ProductBench, covering 15 categories with 200 products each. Across four search-capable models, CORE promotes target products into the top five over 90 percent of the time. It beats rival manipulation methods while keeping the text fluent.
Catalog record
| CREATOR | Haibo Jin, Ruoxi Chen, Peiyan Zhang, Yifeng Luo, Huimin Zeng, Man Luo, Haohan Wang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 85 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-03608