Artificial Text Detection via Examining the Topology of Attention Maps
arXiv:2109.04825 · doi:10.18653/v1/2021.emnlp-main.50
Abstract
The impressive capabilities of recent generative models to create texts that are challenging to distinguish from the human-written ones can be misused for generating fake news, product reviews, and even abusive content. Despite the prominent performance of existing methods for artificial text detection, they still lack interpretability and robustness towards unseen models. To this end, we propose three novel types of interpretable topological features for this task based on Topological Data Analysis (TDA) which is currently understudied in the field of NLP. We empirically show that the features derived from the BERT model outperform count- and neural-based baselines up to 10\% on three common datasets, and tend to be the most robust towards unseen GPT-style generation models as opposed to existing methods. The probing analysis of the features reveals their sensitivity to the surface and syntactic properties. The results demonstrate that TDA is a promising line with respect to NLP tasks, specifically the ones that incorporate surface and structural information.
Accepted to EMNLP 2021
References in corpus (3)
Cited by in corpus (10)
- Intrinsic Dimension Estimation for Robust Detection of AI-Generated Texts
- Acceptability Judgements via Examining the Topology of Attention Maps
- Findings of the The RuATD Shared Task 2022 on Artificial Text Detection in Russian
- An Empirical Study of AI Generated Text Detection Tools
- Can BERT eat RuCoLA? Topological Data Analysis to Explain
- Beyond Simple Averaging: Improving NLP Ensemble Performance with Topological-Data-Analysis-Based Weighting
- Better Call Claude: Can LLMs Detect Changes of Writing Style?
- AI-generated text boundary detection with RoFT
- The Density of Cross-Persistence Diagrams and Its Applications
- Abstraction, Reasoning and Deep Learning: A Study of the "Look and Say" Sequence