GRADA: Graph-based Reranking against Adversarial Documents Attack
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908544573374464 |
|---|---|
| author | Zheng, Jingjie Gema, Aryo Pradipta Hong, Giwon He, Xuanli Minervini, Pasquale Sun, Youcheng Xu, Qiongkai |
| author_facet | Zheng, Jingjie Gema, Aryo Pradipta Hong, Giwon He, Xuanli Minervini, Pasquale Sun, Youcheng Xu, Qiongkai |
| contents | Retrieval Augmented Generation (RAG) frameworks improve the accuracy of large language models (LLMs) by integrating external knowledge from retrieved documents, thereby overcoming the limitations of models' static intrinsic knowledge. However, these systems are susceptible to adversarial attacks that manipulate the retrieval process by introducing documents that are adversarial yet semantically similar to the query. Notably, while these adversarial documents resemble the query, they exhibit weak similarity to benign documents in the retrieval set. Thus, we propose a simple yet effective Graph-based Reranking against Adversarial Document Attacks (GRADA) framework aiming at preserving retrieval quality while significantly reducing the success of adversaries. Our study evaluates the effectiveness of our approach through experiments conducted on five LLMs: GPT-3.5-Turbo, GPT-4o, Llama3.1-8b, Llama3.1-70b, and Qwen2.5-7b. We use three datasets to assess performance, with results from the Natural Questions dataset demonstrating up to an 80% reduction in attack success rates while maintaining minimal loss in accuracy. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_07546 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | GRADA: Graph-based Reranking against Adversarial Documents Attack Zheng, Jingjie Gema, Aryo Pradipta Hong, Giwon He, Xuanli Minervini, Pasquale Sun, Youcheng Xu, Qiongkai Information Retrieval Artificial Intelligence Retrieval Augmented Generation (RAG) frameworks improve the accuracy of large language models (LLMs) by integrating external knowledge from retrieved documents, thereby overcoming the limitations of models' static intrinsic knowledge. However, these systems are susceptible to adversarial attacks that manipulate the retrieval process by introducing documents that are adversarial yet semantically similar to the query. Notably, while these adversarial documents resemble the query, they exhibit weak similarity to benign documents in the retrieval set. Thus, we propose a simple yet effective Graph-based Reranking against Adversarial Document Attacks (GRADA) framework aiming at preserving retrieval quality while significantly reducing the success of adversaries. Our study evaluates the effectiveness of our approach through experiments conducted on five LLMs: GPT-3.5-Turbo, GPT-4o, Llama3.1-8b, Llama3.1-70b, and Qwen2.5-7b. We use three datasets to assess performance, with results from the Natural Questions dataset demonstrating up to an 80% reduction in attack success rates while maintaining minimal loss in accuracy. |
| title | GRADA: Graph-based Reranking against Adversarial Documents Attack |
| topic | Information Retrieval Artificial Intelligence |
| url | https://arxiv.org/abs/2505.07546 |