Establish a controlled cross-framework robustness comparison

Establish a controlled robustness comparison between GRICS and other Graph-RAG systems using identical knowledge graphs, ontology structures, attack settings, reasoning tasks, and evaluation protocols, so that direct quantitative superiority can be determined.

Background

The reported comparison with PoisonRAG and GRAPPOISON uses different knowledge graphs, graph-construction procedures, adversarial implementations, and evaluation protocols. Consequently, the reported Attack Success Rate and Tokens per Query values provide qualitative context rather than a controlled head-to-head evaluation.

A common experimental setting is required to determine whether GRICS is quantitatively more robust than competing Graph-RAG systems rather than merely different under incomparable conditions.

References

Consequently, the reported ASR and TPQ values should be interpreted as providing qualitative insight into the robustness characteristics of graph-grounded retrieval systems rather than establishing direct quantitative superiority. A controlled comparison using identical knowledge graphs, attack settings, and evaluation protocols remains an important direction for future work.

— NeuroGraph: An AI Graph-Driven Neuro-Symbolic Framework for Explainable Threat Reasoning in Advanced Manufacturing  (2609.00604 - Nandiya et al., 1 Sep 2026) in Evaluation, subsection “Robustness Against Adversarial Attacks”; Discussion, subsection “Limitations and Open Research Questions,” subsection “Limitations”