Researchers have developed a novel method for attributing authorship of text generated by large language models (LLMs) by analyzing their reasoning structures. This approach utilizes reasoning graphs extracted via an argument mining pipeline and processed by a graph neural network. The new method demonstrates significantly improved robustness and generalization compared to traditional baselines, outperforming them by up to 27 percentage points against obfuscation techniques like paraphrasing and backtranslation. AI
IMPACT This research could lead to more reliable methods for detecting AI-generated content, crucial for maintaining trust and integrity in digital communication.
RANK_REASON The cluster contains a research paper detailing a new method for LLM authorship attribution.
- argument mining
- arXiv
- CatalyzeX
- DagsHub
- Gotit.pub
- graph neural network
- Hugging Face
- large language models
- Longformer: The Long-Document Transformer
- reasoning graphs
- ScienceCast
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
