Adnan KhanAdnanProduct & AI
San Francisco Bay Area
Get in touch

Source note2 min read

Show Me How You Reason and I'll Tell You Who You Are: Reasoning Graphs for Robust LLM Authorship Attribution

Source

arXiv

arxiv.org/abs/2607.14905v1

Read the source

Source

Plain-Language Brief

The paper proposes a new method for detecting text generated by large language models (LLMs) and attributing authorship to them. The method goes beyond just analyzing surface-level linguistic features and instead focuses on reasoning structures in the texts, using a graph neural network approach. This approach is shown to be more robust and generalizable than a traditional Longformer baseline, performing better under paraphrasing and backtranslation attacks, as well as when evaluated on unseen model versions.

What the Document Says

  • The paper discusses the issue of detecting text generated by large language models (LLMs) and attributing authorship to them due to their increasing use in various contexts.
  • Prior work primarily focuses on surface-level linguistic features, which are susceptible to paraphrasing and obfuscation techniques.
  • The proposed method in this paper goes beyond the linguistic surface, focusing on reasoning structures in LLM-generated texts to capture more complex signals of authorship.
  • A graph neural network approach is used, leveraging reasoning graphs extracted by an argument mining pipeline.
  • This approach shows improved robustness and generalization over a traditional Longformer baseline.
  • The approach outperforms the baseline by up to 27 percentage points under obfuscation attacks such as paraphrasing and backtranslation, and by 19 percentage points when evaluated on unseen model versions.

Key Concepts Mentioned

  • large language models
  • authorship attribution
  • reasoning structures
  • graph neural networks
  • argument mining
  • paraphrasing
  • backtranslation

Claims Made by the Document

  • The proposed method improves robustness and generalization over a traditional Longformer baseline.
  • The approach outperforms the baseline under obfuscation attacks.

Evidence Used

  • improved results when compared to a traditional Longformer baseline
  • outperforming the baseline by specific percentage points under certain conditions

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

Given the current trend to employ large language models (LLMs) in almost any imaginable context, LLM-generated text detection and authorship attribution have become a pressing issue. Prior work has primarily focused on surface-level linguistic features, an approach shown to be susceptible to paraphrasing and other obfuscation techniques. In this paper, we go beyond the linguistic surface, extracting and analysing reasoning structures in LLM-generated texts with the goal of capturing more complex signals of LLM authorship. We propose a graph neural network approach that leverages reasoning graphs extracted by an argument mining pipeline, demonstrating improved robustness and generalisation over a traditional Longformer baseline. Our approach outperforms the baseline by up to 27 percentage points under the obfuscation attacks such as paraphrasing and backtranslation, and 19 percentage points when evaluated on the texts generated by the unseen model versions, simulating real-world conditions in which new LLM versions are continuously released.