GRADA: Graph-based Reranking against Adversarial Documents Attack

TL;DR

GRADA is a graph-based reranking method designed to defend retrieval-augmented language models against adversarial document attacks, significantly reducing attack success while maintaining high retrieval accuracy.

Contribution

This paper introduces GRADA, a novel graph-based reranking framework that enhances the robustness of RAG systems against adversarial document manipulations.

Findings

Up to 80% reduction in attack success rates.

Maintains minimal loss in retrieval accuracy.

Effective across multiple large language models.

Abstract

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.