Datasets:

LThurnay's picture
Update README.md
912e27a verified
|
raw
history blame
1.85 kB
metadata
task_categories:
  - text-classification
language:
  - de
tags:
  - legal
  - information obligation
pretty_name: Information obligations in the Ausrian Legal Information System (RIS)
size_categories:
  - n<1K

Federal consolidated section* documents from the Austrian Legal Information System (RIS), classified whether they prescribe information obligation (positive) or not (negative), as defined by the Austrian Corporate Service Portal Act (USPG) § 1 Abs. 1.

  • full.txt: the full corpus in fasttext format.
  • full_with_nor.txt: the full corpus in fasttext format, where the first token (starting with NOR) is the unique ID of the document as can be found by in the above linked database.
  • train.txt, dev.txt, and test.txt are the 80:10:10 split of the full corpus to train, validate, and test it, as was used to train the uwk-cegov/information_obligation_classifier model.
label training docs training words valid. docs valid. words test docs test words total docs total words
positive 155 79.9% 100 927 74.7% 19 9.8% 8 791 6.5%
negative 462 79.9% 456 300 75.5% 57 9.9% 67 627 11.2%
total 617 79.9% 557 227 75.4% 76 9.8% 76 418 10.3%

* Consolidation means that all amendments and corrections subsequently promulgated have been incorporated into a piece of legislation. These documents are for information purposes only and are therefore not legally binding. The documentation unit (a document) is a section (Paragraf) or an article or an annex to a legal provision.