ADAPT researchers Abigail Walsh (DCU), Carla Parra (DCU) and Alfredo Maldonado (TCD) are involved in preparing a large, multilingual corpus which will be used in the second edition of a PARSEME Shared Task on automatic detection of verbal multiword expressions (MWEs). The corpus will be annotated with verbal multiword expressions, made using a common set of annotation guidelines for all languages. Carla and Abigail are co-ordinating the generation of the Spanish and English annotations respectively.
The PARSEME Shared Task is a large, international, collaboration with the aim of creating a data set in multiple languages from very different language families. The first edition of the Shared Task ran in 2017 and utilised existing Natural Language Processing (NLP) tools to automatically identify verbal MWEs. Edition 1.0 involved 7 participating systems, spanning 18 languages. The next shared task, edition 1.1, will include new languages, enhanced guidelines, larger corpora, and the possibility to align with manually annotated treebanks.
The PARSEME Shared Task 1.1 will be collocated at the upcoming LAW-MWE-CxG workshop at the International Conference on Computational Linguistics (COLING) in August 2018.
Share this article: