Multilingual Approaches to NLP

Multilingual NLP 2021


Artificial Intelligence





Special Issue on "Multilingual Approaches to NLP"
*Aims and Scope*
Multilingual approaches to natural language processing (NLP) have become increasingly popular with the field’s growing awareness of the limitations of monolingual approaches, and the realisationthat a single language can never be representative for the whole world’s linguistic diversity. One constant obstacle to multilingual NLP, is the access to sufficient labelled data in low-resource languages. This is partially alleviated by multilingual resources such as the Universal Dependencies, and Unimorph. A growing body of work has focused on transfer learning methods which often use data from relatively high-resource languages, for low-resource ones. This approach is crucial to the success of NLP for low-resource languages, as it is unfeasible to obtain labelled data for all languages in the world. Furthermore, even high-resource languages may benefit from multilingual transfer from other languages.
*Main Topics*
-Multilingual NLP
-Language-independent training, architecture design, and hyperparameter tuning. -Integration of typological features in multilingual learning
-Typologically inspired NLP architectures
-Cross-lingual transfer
-Low-resource NLP
-Linguistic Diversity and Fairness
-Interpretability of Multilingual Models
-Evaluation of language-independent methods
-Adaptation of monolingual methods to cross-lingual settings
-Construction / annotation of multilingual resources
-Techniques for simultaneous modelling of several languages
*How to Submit*
Please visit journal author guideline at: https://www.atlantis-press.com/journals/nlpr/author-guidelines
more information please contact Yanhua.li@atlantis-press.com