Universal Dependencies
Brazilian Festival

The Universal Dependencies Brazilian Festival (UDFest-BR) will take place with the 15th edition of the International Conference on the Computational Processing of Portuguese (PROPOR 2022). It seeks to become a forum in which researchers involved with the study and application of the Universal Dependencies (UD) model, along with its adaptation to Portuguese, can meet and discuss best practices and strategies, as well as problems and solutions related to this topic.

UD is a cross-language international model for grammatical annotation (morphosyntactic tags, lexical/morphosyntactic features and syntactic dependencies), which results from the efforts of an open community of hundreds of researchers. Building upon a common framework, the model captures idiosyncrasies and similarities between different languages, aiming at fostering contrastive linguistic studies and the development of multilingual technologies for Natural Language Processing (NLP).

In recent years, this model has drawn considerable attention from the NLP community. This growing interest can be owed mainly to the fact that UD allows for the development of fairly precise taggers and parsers to a number of languages, based on the exploitation of syntactically annotated corpora (treebanks) and some Machine Learning algorithms. Currently, we find around 200 UD annotated corpora available in more than 100 languages.

The popularity of the UD model also results from the fact that dependency analysis seems to be useful to various NLP applications (such as automatic translation and information extraction), since it more transparently codifies predicate-argument structures. Another reason to its prominence lies in the high adequacy of dependency models (when compared to traditional constituent structures) to languages with more flexible word ordering.

Although there can be found some annotated corpora available in Portuguese, computational linguistic researches involving UD and this language are still incipient. As such, the existence of a workshop dedicated to this topic will certainly help fill in this gap, giving more visibility to current efforts in this area as well as fostering new initiatives and collaborations.