The attachment scores for cross-lingual models even surpass the fully supervised models trained on the target language treebank . Croatian is the most difficult language in the test set and the improvements over the baseline are rather modest . Norwegian works best with information coming from Swedish whereas Danish contributes surprisingly little to the task of cross-linking dependency parsing at VarDial 2017 . In particular, Slovak seems to work well with information from the Czech treebank, which is in line with related work, according to the University of Helsinki . The attachment score is higher than those trained on fully supervised language treebanks .

Links: PDF - Abstract

Code :


Keywords : language - cross - - helsinki - related -

Leave a Reply

Your email address will not be published. Required fields are marked *


Enjoy this blog? Please spread the word :)