Datasets:

---

danish-foundation-models

/

danish-dynaword

like 8

Follow

Danish Foundation Models 40

All danish digitized books in the period period 1750 - 1930 fetched from https://soeg.kb.dk/

Obtained from:

https://soeg.kb.dk/discovery/search?query=any,contains,*%20*,AND&pfilter=lang,exact,dan,AND&pfilter=rtype,exact,books,AND&pfilter=dr_s,exact,17500101,AND&pfilter=dr_e,exact,19301231,AND&tab=Everything&search_scope=MyInst_and_CI&sortby=rank&vid=45KBDK_KGL:KGL&mfacet=tlevel,include,online_resources,1&lang=da&mode=advanced&offset=0

Currently a dump is available on UCloud under the project GoldenMatrix

kb-books/raw

Here will need to do the following filtering:

• 1. Convert it to a dataframe on the metadata in the JSON + .pdf + text extract from the pdf (it is already OCR'ed but badly)
• 2. Based on the metadata we need to filter out all authors who have been dead for more than 70 years
• 3. Once that is done (and review by Per) we can publish that dataset to the Huggingface, under CHCAA. It should be such that other can re-OCR the data
• 4. From that dataset, we can then add a the text data to dynaword