Lune

ACL Findings2023

Are Synonym Substitution Attacks Really Synonym Substitution Attacks?

David Cheng-Han Chiang, Hung-yi Lee

Abstract

In this paper, we explore the following question: Are synonym substitution attacks really synonym substitution attacks (SSAs)? We approach this question by examining how SSAs replace words in the original sentence and show that there are still unresolved obstacles that make current SSAs generate invalid adversarial samples. We reveal that four widely used word substitution methods generate a large fraction of invalid substitution words that are ungrammatical or do not preserve the original sentence's semantics. Next, we show that the semantic and grammatical constraints used in SSAs for detecting invalid word replacements are highly insufficient in detecting invalid adversarial samples.

Ask Lune

Ask anything. The answer quotes the paper.

Lune reads all of it and the papers around it, then shows you the passages behind every answer.

Questions to start from

Lune answers from

the full paper · 3 top-tier papers that cite it · related work

Get the answer

Free to start. No credit card required.

Cited by top-tier papers3

  1. Automatically Generated Definitions and their utility for Modeling Word MeaningFrancesco Periti, David Alfter, Nina TahmasebiEMNLP 2024 · 2 citations
  2. YESciEval: Robust LLM-as-a-Judge for Scientific Question AnsweringJennifer D'Souza, Hamed Babaei Giglou, Quentin MünchACL 2025
  3. Precise Verification of Transformers Through ReLU-Catalyzed Abstraction RefinementHengjie Liu, Zhenya Zhang, Jianjun ZhaoCAV 2026

Builds on8

  1. Is BERT Really Robust? A Strong Baseline for Natural Language Attack on Text Classification and EntailmentDi Jin, Zhijing Jin, Joey Tianyi Zhou, Peter SzolovitsAAAI 2020 · 1,333 citations
  2. BERT-ATTACK: Adversarial Attack Against BERT Using BERTLinyang Li, Ruotian Ma, Qipeng Guo, Xiangyang Xue et al.EMNLP 2020 · 529 citations
  3. Towards Improving Adversarial Training of NLP ModelsJin Yong Yoo, Yanjun QiEMNLP Findings 2021 · 94 citations
  4. Reevaluating Adversarial Examples in Natural LanguageJohn X. Morris, Eli Lifland, Jack Lanchantin, Yangfeng Ji et al.EMNLP Findings 2020 · 93 citations
  5. It's Morphin' Time! Combating Linguistic Discrimination with Inflectional PerturbationsSamson Tan, Shafiq R. Joty, Min-Yen Kan, Richard SocherACL 2020 · 88 citations

Related papers

  1. Evaluating the Validity of Word-level Adversarial Attacks with Large Language ModelsHuichi Zhou, Zhaoyang Wang, Hongtao Wang, Dongping Chen et al.ACL Findings 2024 · 6 citations
  2. Bigram and Unigram Based Text Attack via Adaptive Monotonic Heuristic SearchXinghao Yang, Weifeng Liu, James Bailey, Dacheng Tao et al.AAAI 2021 · 12 citations
  3. SSPAttack: A Simple and Sweet Paradigm for Black-Box Hard-Label Textual Adversarial AttackHan Liu, Zhi Xu, Xiaotong Zhang, Xiaoming Xu et al.AAAI 2023 · 31 citations
  4. Contrasting Human- and Machine-Generated Word-Level Adversarial Examples for Text ClassificationMaximilian Mozes, Max Bartolo, Pontus Stenetorp, Bennett Kleinberg et al.EMNLP 2021 · 8 citations
  5. CheckHARD: Checking Hard Labels for Adversarial Text Detection, Prediction Correction, and Perturbed Word SuggestionHoang-Quoc Nguyen-Son, Huy Quang Ung, Seira Hidano, Kazuhide Fukushima et al.EMNLP Findings 2022 · 3 citations
Dusk over the sea between two cliffs drawn in fine vertical lines