75 Languages, 1 Model: Parsing Universal Dependencies Universally
This provides a unified parsing solution for many languages, particularly benefiting low-resource ones through cross-linguistic transfer, though it is incremental as it builds on existing BERT and UD frameworks.
The authors tackled multilingual dependency parsing across 75 languages by fine-tuning a multilingual BERT model on all Universal Dependencies treebanks, achieving state-of-the-art scores in UPOS, UFeats, Lemmas, UAS, and LAS without language-specific components.
We present UDify, a multilingual multi-task model capable of accurately predicting universal part-of-speech, morphological features, lemmas, and dependency trees simultaneously for all 124 Universal Dependencies treebanks across 75 languages. By leveraging a multilingual BERT self-attention model pretrained on 104 languages, we found that fine-tuning it on all datasets concatenated together with simple softmax classifiers for each UD task can result in state-of-the-art UPOS, UFeats, Lemmas, UAS, and LAS scores, without requiring any recurrent or language-specific components. We evaluate UDify for multilingual learning, showing that low-resource languages benefit the most from cross-linguistic annotations. We also evaluate for zero-shot learning, with results suggesting that multilingual training provides strong UD predictions even for languages that neither UDify nor BERT have ever been trained on. Code for UDify is available at https://github.com/hyperparticle/udify.