CoolFace
Datasetpublic

V4ldeLund/faroese-taks-fo

TAKS (taks_fo) Status: Standalone website collection. It is not part of Faroese Dynaword. About this data Tax and customs laws, circulars, guidance, forms, HTML pages, PDFs, DOCX/CSV files and linked Kærustovnur decisions. Source: TAKS Size Measure Value Documents 1,178 Characters 13,463,572 Tokens 5,127,975 Possible official documents 205 Tokens in possible official documents 765,561 File type Documents HTML 852… See the full description on the dataset page: https://huggingface.co/datasets/V4ldeLund/faroese-taks-fo.

sourceHugging Faceupdated 29d agoView on Hugging Face
0likes64downloads
Dataset Card

TAKS (taks_fo)

Status: Standalone website collection. It is not part of Faroese Dynaword.

About this data

Tax and customs laws, circulars, guidance, forms, HTML pages, PDFs, DOCX/CSV files and linked Kærustovnur decisions.

Source: TAKS

Size

MeasureValue
Documents1,178
Characters13,463,572
Tokens5,127,975
Possible official documents205
Tokens in possible official documents765,561
File typeDocuments
HTML852
PDF298
DOCX25
CSV3
plain text0

The token count uses the AI-Sweden-Models/Llama-3-8B-instruct tokenizer. It counts all collected text, including other languages and repeated website text.

Files

  • —rows.parquet contains the collected documents.
  • —parser/ contains the crawler, site settings and Python requirements.

Crawl details

  • —Crawl started: 2026-08-16
  • —Crawl completed: 2026-08-16

The crawler used public sitemaps and links from the website. It stayed within the websites and file hosts listed in parser/sites.csv and followed robots.txt.

Text was taken from web pages, text-based PDFs, Word files and CSV files. Scanned PDFs and images were not read with OCR. Very short pages and exact duplicate texts were left out.

Columns

ColumnMeaning
source_site, source_homepageWebsite name and homepage
source_url, requested_urlFinal URL and the URL first requested
title, text, content_typeDocument title, extracted text and file type
retrieved_atWhen the document was downloaded
text_sha256, text_charactersText hash and character count
rights_status, rights_basisAutomatic rights label and why it was assigned
extraction_source_url, source_page_urlExtra source links when available

Rights

Some laws, decisions, notices and similar public documents may be outside copyright under section 9 of the Faroese Copyright Act.

likely_official_document only means that the title or URL matched a list of words. It is not a legal classification.

Note for this site

Laws, circulars, official tax guidance and final appeal decisions are the clearest candidates. Use the Logir version of a law when available. Service pages, forms and outside attachments may still be copyrighted.

Known limits

  • —The crawl may not cover every page or old document.
  • —The text has not been checked by hand or filtered by language.
  • —Menus, headers and other repeated website text may remain.
  • —Text from PDFs may appear in the wrong order.
  • —Scanned PDFs and text inside images are missing.
  • —Some pages may contain personal information.

Parser

The crawler used for this website is included in the parser folder. The copy of sites.csv contains only this website.

bash
cd parser
python -m pip install -r requirements.txt
python scrape.py --site taks_fo