nld.rst.nldt
Dutch Discourse Treebank (NLDT)
Description
Citation: Gisela Redeker, Ildikó Berzlánovich, Nynke van der Vliet, Gosse Bouma and Markus Egg (2012). Multi-Layer Discourse Annotation of a Dutch Text Corpus. In: Proceedings of LREC 2012. Istanbul, Turkey, 2820-2825.
NLDT is a corpus of 80 Dutch texts from expository texts (encyclopedias and science news web pages) and persuasive genres (fund-raising letters and commercial advertisements), which were annotated for rhetorical and genre-specific discourse structure, and lexical cohesion with the goal of creating a gold standard for further research. The annotations are based on a segmentation of the texts into elementary discourse units that takes into account cues from syntax and punctuation.
DISRPT 2021 shared task information
For the DISRPT 2021 shared task, train/dev/test partitions were balanced for genre. Syntactic parses in UD were produced using Stanza. The data contains no split EDUs, which were prevented in a handful of cases through some reordering of clauses in the original RST data. These segment orders were retained in the 2021 shared task data.