Simpler Becomes Harder: Do LLMs Exhibit a Coherent Behavior on Simplified Corpora?

Miriam Anschütz, Edoardo Mosca, Georg Groh


Abstract
Text simplification seeks to improve readability while retaining the original content and meaning. Our study investigates whether pre-trained classifiers also maintain such coherence by comparing their predictions on both original and simplified inputs. We conduct experiments using 11 pre-trained models, including BERT and OpenAI’s GPT 3.5, across six datasets spanning three languages. Additionally, we conduct a detailed analysis of the correlation between prediction change rates and simplification types/strengths. Our findings reveal alarming inconsistencies across all languages and models. If not promptly addressed, simplified inputs can be easily exploited to craft zero-iteration model-agnostic adversarial attacks with success rates of up to 50%.
Anthology ID:
2024.determit-1.17
Volume:
Proceedings of the Workshop on DeTermIt! Evaluating Text Difficulty in a Multilingual Context @ LREC-COLING 2024
Month:
May
Year:
2024
Address:
Torino, Italia
Editors:
Giorgio Maria Di Nunzio, Federica Vezzani, Liana Ermakova, Hosein Azarbonyad, Jaap Kamps
Venues:
DeTermIt | WS
SIG:
Publisher:
ELRA and ICCL
Note:
Pages:
185–195
Language:
URL:
https://aclanthology.org/2024.determit-1.17
DOI:
Bibkey:
Cite (ACL):
Miriam Anschütz, Edoardo Mosca, and Georg Groh. 2024. Simpler Becomes Harder: Do LLMs Exhibit a Coherent Behavior on Simplified Corpora?. In Proceedings of the Workshop on DeTermIt! Evaluating Text Difficulty in a Multilingual Context @ LREC-COLING 2024, pages 185–195, Torino, Italia. ELRA and ICCL.
Cite (Informal):
Simpler Becomes Harder: Do LLMs Exhibit a Coherent Behavior on Simplified Corpora? (Anschütz et al., DeTermIt-WS 2024)
Copy Citation:
PDF:
https://preview.aclanthology.org/add_acl24_videos/2024.determit-1.17.pdf