German compound splitting dataset deu

The compounds that were used in Ma et al (2016) paper entitled "Letter Sequence Labeling for Compound Splitting". It contains both two-constituent and multi-constituent compounds. As standard evaluation also involves non-compounds, the data also include non-compounds that we used. The data are organized into the exact same training/test/development split as in the paper. eng

f1a5caec-3529-4b1c-9473-b23f073ee8b6

https://doi.org/10.57754/FDAT.ydh6f-rgv47

2025-02-04T10:52:47.125Z

lexical_resource

8cefa5dd-f5fb-4527-8acb-88cc6824eb48