Topological Attribution Distance (TAD): Revealing Segment-Level RAG Influence on LLM Output Geometry for Incident Log Analysis

Source

• Source name: arXiv
• Source kind: paper
• Authors / organization: Reza Fayyazi, Michael Zuzak, Shanchieh Jay Yang
• Venue: arXiv: cs.CR
• Published: 2026-08-17
• Collected: 2026-08-18
• URL: http://arxiv.org/abs/2608.16775v1
• PDF: https://arxiv.org/pdf/2608.16775v1
• DOI: UNKNOWN
• Registry group: academic_apis_and_preprints
• Evidence role: canonical_academic_index
• Language: en
• Country / region: UNKNOWN

Plain-Language Brief

Researchers propose a method called Topological Attribution Distance (TAD) to help identify which parts of cyber incident logs are critical in influencing the outputs of Large Language Models (LLMs) used in cybersecurity operations. This could help establish trust in the generated outputs and provide evidence verification.

What the Document Says

• The research focuses on Large Language Models (LLMs) in cybersecurity operations for rapid decision-making against threats.
• Trust is essential in the LLM's generated outputs, especially with Agentic AI integration.
• A robust evidence attribution and provenance tracking technique are necessary to trace the origins of model generations.
• Existing methods struggle to distinguish complex and highly similar evidence sources like cyber incident logs.
• The main contribution is the proposal of Topological Attribution Distance (TAD) to characterize and capture the global geometric shape of an output and its changes against retrieved logs.
• TAD uses segment-level ablation attribution to investigate actual cyberattack incident logs.
• TAD can find the most attributed logs on LLM outputs in an adaptive manner.
• This could provide an explainable and trustworthy tracing based on each LLM's hidden state.

Key Concepts Mentioned

• Large Language Models (LLMs)
• Cybersecurity
• Agentic AI
• Evidence attribution
• Provenance tracking
• Topological Attribution Distance (TAD)
• Cyberattack incident logs
• Hidden state

Claims Made by the Document

• Existing methods struggle to distinguish complex and highly similar evidence sources.
• TAD can find the most attributed logs on LLM outputs.

Evidence Used

• LLMs are increasingly being deployed in cybersecurity operations... However, there is a main criteria that must be met when using LLMs in cybersecurity...

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

Large Language Models (LLMs) are increasingly being deployed in cybersecurity operations to assist cybersecurity analysts with rapid decision-making against emerging threats. However, there is a main criteria that must be met when using LLMs in cybersecurity, that is, trust in the generated outputs. As Agentic AI is integrated into operational systems, a robust evidence attribution and provenance tracking technique is essential to trace the origins of model generations. When autonomous agents make a decision (right or wrong), the ability to trace back through the decision chain is critical, as without it, teams cannot identify which segment of the data caused the model generation. Existing methods often struggle to distinguish among complex and highly similar evidence sources, such as cyber incident logs. This reveals a key gap: current approaches do not adequately capture the holistic geometric relationship between the retrieved evidence and the generated response for reliable evidence verification. To bridge this gap, we propose Topological Attribution Distance (TAD), inspired by Topology, to characterize and capture the global geometric shape of an output and its changes against its retrieved logs. In other words, if the embeddings of a specific source log drastically changes the geometry of the model's response in the embedding space, this suggests that such log is a critical source for the model's generated response. Therefore, TAD is powered by segment-level ablation attribution to investigate incident logs of an actual cyberattack. We demonstrate how TAD finds the most attributed logs on LLM outputs in an adaptive manner. This can provide an explainable and trustworthy tracing based on each LLM's hidden state to understand how geometrically different retrieved logs influence the model generation, and provide evidence verification in cybersecurity and Agentic-AI workflows.