Multilingual Part-of-Speech Tagging with Bidirectional Long Short-Term Memory Models and Auxiliary Loss
This work addresses the robustness and performance of bi-LSTMs for multilingual POS tagging, which is an incremental improvement with practical implications for NLP applications in diverse languages.
The paper tackled the problem of understanding bi-LSTM networks for multilingual POS tagging by evaluating their sensitivity to input representations, languages, data size, and label noise, and introduced a novel bi-LSTM model with an auxiliary loss for rare words, achieving state-of-the-art performance across 22 languages, especially for morphologically complex ones.
Bidirectional long short-term memory (bi-LSTM) networks have recently proven successful for various NLP sequence modeling tasks, but little is known about their reliance to input representations, target languages, data set size, and label noise. We address these issues and evaluate bi-LSTMs with word, character, and unicode byte embeddings for POS tagging. We compare bi-LSTMs to traditional POS taggers across languages and data sizes. We also present a novel bi-LSTM model, which combines the POS tagging loss function with an auxiliary loss function that accounts for rare words. The model obtains state-of-the-art performance across 22 languages, and works especially well for morphologically complex languages. Our analysis suggests that bi-LSTMs are less sensitive to training data size and label corruptions (at small noise levels) than previously assumed.