KERAG: Knowledge-Enhanced Retrieval-Augmented Generation for Advanced Question Answering
Yushi Sun, Kai Sun, Yifan Ethan Xu, Xiao Yang, Xin Luna Dong, Nan Tang, Lei Chen · 2025
Read KERAG: Knowledge-Enhanced Retrieval-Augmented Generation for Advanced Question Answering by Yushi Sun, Kai Sun, Yifan Ethan Xu, Xiao Yang, Xin Luna Dong, Nan Tang, Lei Chen (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-2509-04716.txt.
Description
Answering hard questions from a knowledge graph often fails because retrieval is too narrow. Yushi Sun, lead author with six co-authors, present KERAG, a retrieval-augmented generation pipeline, in a 2025 paper. Knowledge graphs store facts as linked entities, offering precise material for question answering.
KERAG retrieves a broader subgraph likely to contain relevant information, then filters and summarises it. Fine-tuned language models reason step by step over those subgraphs, cutting noise. Experiments show gains over leading methods on both simple and complex questions. The authors frame the design as a way to raise coverage without losing answer quality.
Catalog record
| CREATOR | Yushi Sun, Kai Sun, Yifan Ethan Xu, Xiao Yang, Xin Luna Dong, Nan Tang, Lei Chen |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 94 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2509-04716