CharacterBERT: Reconciling ELMo and BERT for Word-Level Open-Vocabulary Representations From Characters
This addresses the issue of domain-specific tokenization for researchers and practitioners in fields like medicine, though it is an incremental improvement over BERT.
The authors tackled the problem of BERT's reliance on wordpiece tokenization, which is unsuitable for specialized domains like medicine, by proposing CharacterBERT, a variant that uses character-level representations; they showed it improves BERT's performance on medical tasks, producing robust, word-level, open-vocabulary representations.
Due to the compelling improvements brought by BERT, many recent representation models adopted the Transformer architecture as their main building block, consequently inheriting the wordpiece tokenization system despite it not being intrinsically linked to the notion of Transformers. While this system is thought to achieve a good balance between the flexibility of characters and the efficiency of full words, using predefined wordpiece vocabularies from the general domain is not always suitable, especially when building models for specialized domains (e.g., the medical domain). Moreover, adopting a wordpiece tokenization shifts the focus from the word level to the subword level, making the models conceptually more complex and arguably less convenient in practice. For these reasons, we propose CharacterBERT, a new variant of BERT that drops the wordpiece system altogether and uses a Character-CNN module instead to represent entire words by consulting their characters. We show that this new model improves the performance of BERT on a variety of medical domain tasks while at the same time producing robust, word-level and open-vocabulary representations.