TextFooler for Chinese, method proposed by "Is BERT Really Robust? A Strong Baseline for Natural Language Attack on Text Classification and Entailment" https://arxiv.org/abs/1907.11932
- Use spacy to find topN similar words
- Use Universal Sentence Encoder for Chinese to get sentence embedding