Dataset and scripts for the DRASTIC corpus of DRS-annotated texts.
The repository has the following structure:
├── data
│ ├── drs-annotation
│ │ ├── anaphora-resolution
│ │ │ ├── dvorak
│ │ │ ├── marbles
│ │ │ ├── nida
│ │ │ └── short-texts
│ │ └── no-anaphora-resolution
│ │ ├── dvorak
│ │ ├── marbles
│ │ ├── nida
│ │ └── short-texts
│ └── ud-sources
│ ├── dvorak
│ ├── marbles
│ ├── nida
│ └── short-texts
└── scripts
data contains the drs-annotation, in a clausal format, as well as the corresponding ud-sources from the GUM corpus. The semantic annotations are given in two versions: one with sentence-internal anaphora resolved (anaphora-resolution) and one without (no-anaphora-resolution). Within each directory, the texts are divided by sub-corpus, and named for their corresponding UD sent_ids (for details, see Haug et al. 2023, referenced below).
scripts contains a script (flatten_clause_notation.py) which will 'flatten' PMB-style DRSs into our simplified format. We also provide a shell script (flatten_clause_notation_in_batch.sh) to run this on multiple files at once.
If you use this data, please cite the following paper:
Haug, Dag T. T., Jamie Y. Findlay and Ahmet Yıldırım. 2023. The long and the short of it: DRASTIC, a semantically annotated dataset containing sentences of more natural length. In Proceedings of the 4th International Workshop on Designing Meaning Representations (DMR 2023), 89–98. Association for Computational Linguistics.
@inproceedings{haug_etal:drastic,
title = {The long and the short of it: \textsc{drastic}, a semantically annotated dataset containing sentences of more natural length},
year = {2023},
author = {Dag T. T. Haug and Jamie Y. Findlay and Ahmet Y\i{}ld\i{}r\i{}m},
booktitle = {{Proceedings of the 4th International Workshop on Designing Meaning Representations (DMR 2023)}},
pages = {89--98},
publisher = {Association for Computational Linguistics},
url = {https://aclanthology.org/2023.dmr-1.9}
}
This data is licensed under a Creative Commons Attribution 4.0 International License.
