@inproceedings{aguilar-etal-2021-char2subword-extending,
    title = "{C}har2{S}ubword: Extending the Subword Embedding Space Using Robust Character Compositionality",
    author = "Aguilar, Gustavo  and
      McCann, Bryan  and
      Niu, Tong  and
      Rajani, Nazneen  and
      Keskar, Nitish Shirish  and
      Solorio, Thamar",
    booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2021",
    month = nov,
    year = "2021",
    address = "Punta Cana, Dominican Republic",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2021.findings-emnlp.141",
    doi = "10.18653/v1/2021.findings-emnlp.141",
    pages = "1640--1651",
    abstract = "Byte-pair encoding (BPE) is a ubiquitous algorithm in the subword tokenization process of language models as it provides multiple benefits. However, this process is solely based on pre-training data statistics, making it hard for the tokenizer to handle infrequent spellings. On the other hand, though robust to misspellings, pure character-level models often lead to unreasonably long sequences and make it harder for the model to learn meaningful words. To alleviate these challenges, we propose a character-based subword module (char2subword) that learns the subword embedding table in pre-trained models like BERT. Our char2subword module builds representations from characters out of the subword vocabulary, and it can be used as a drop-in replacement of the subword embedding table. The module is robust to character-level alterations such as misspellings, word inflection, casing, and punctuation. We integrate it further with BERT through pre-training while keeping BERT transformer parameters fixed{--}and thus, providing a practical method. Finally, we show that incorporating our module to mBERT significantly improves the performance on the social media linguistic code-switching evaluation (LinCE) benchmark.",
}