---
license: mit
---

In this dataset, we present a dataset of 2944 legal references in German law that are manually annotated by law experts. This dataset has 21 properties for each law reference in the dataset, such as _Buch_, _Teil_, _Titel_, _Untertitel_, etc. It also provides the complete text of each law reference in the dataset, along with specific paragraph text mentioned in the law reference.

Paper: [A Dataset of German Legal Reference Annotations](https://scholar.google.com/citations?view_op=view_citation&hl=en&user=c5KToK8AAAAJ&citation_for_view=c5KToK8AAAAJ:9yKSN-GCB0IC)

Please reference our work when using this dataset:

```tex
@inproceedings{10.1145/3594536.3595173,
  author = {Darji, Harshil and Mitrovi\'{c}, Jelena and Granitzer, Michael},
  title = {A Dataset of German Legal Reference Annotations},
  year = {2023},
  isbn = {9798400701979},
  publisher = {Association for Computing Machinery},
  address = {New York, NY, USA},
  url = {https://doi.org/10.1145/3594536.3595173},
  doi = {10.1145/3594536.3595173},
  abstract = {The field of legal Natural Language Processing faces a lot of challenges due to the unavailability of properly structured datasets. One such instance is the need for a dataset that not only separates different parts of legal references, such as an article or paragraph number but also provides information about what a particular legal reference dictates. Having access to such a dataset can provide easy access to researchers working on experiments such as context similarity between law texts and legal cases that refer to a particular law. In this paper, we present a dataset of 2944 legal references in German law that are manually annotated by law experts. This dataset has 21 properties for each law reference in the dataset, such as Buch, Teil, Titel, Untertitel, etc. It also provides the complete text of each law reference in the dataset, along with specific paragraph text mentioned in the law reference. Furthermore, using this dataset together with Open Legal Data, we perform a law reference prediction task to compare the performance between predicting full law reference and only the base law reference.},
  booktitle = {Proceedings of the Nineteenth International Conference on Artificial Intelligence and Law},
  pages = {392–396},
  numpages = {5},
  keywords = {NLP, Law Reference Annotations, Sentence Transformers, Legal Language Processing, Law References, Open Legal Data},
  location = {Braga, Portugal},
  series = {ICAIL '23}
}
```