Enhancing Domain-Specific Retrieval-Augmented Generation: Synthetic Data Generation and Evaluation using Reasoning Models
1 st Aryan Jadon, 2 nd Avinash Patil, 3 rd Shashank Kumar · 2025
Read Enhancing Domain-Specific Retrieval-Augmented Generation: Synthetic Data Generation and Evaluation using Reasoning Models by 1 st Aryan Jadon, 2 nd Avinash Patil, 3 rd Shashank Kumar (2025) 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-2502-15854.txt.
Description
Aryan Jadon and co-authors sharpen retrieval for AI assistants in technical domains with synthetic practice data in a 2025 paper. Assistants that fetch facts stumble on dense filings, medical abstracts, and threat reports. Document-level scores hide exactly where small chunks lose meaning.
New token-level metrics score how much context a chunk keeps versus how much it bloats. Reasoning models then mint question-answer pairs pinned to scattered spans across the three corpora. Tiny chunks proved most precise but missed more, and the best chunk size swung by field. The framework gives builders a tuning bench for domain retrieval.
Catalog record
| CREATOR | 1 st Aryan Jadon, 2 nd Avinash Patil, 3 rd Shashank Kumar |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 37 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-15854