The dataset contains EU legislation documents with manually annotated reporting obligations. Annotation was performed manually by legal experts. Each document was read and annotated by a legal expert only once. Therefore, the might not be consistency of annotations throughout the documents. Reporting obligation annotations may include various texts, but not limited to: definitions, specifications, context, requirements, recitals, exceptions. The dataset can be used to train language models for reporting obligations detection in legislative texts. It can also be used to analyse the complexity and structure of reporting obligations in texts.
