--- language: - sr - hr - sl - bs pretty_name: ZNANJE size_categories: - 1B

Skup naučnih publikacija na Južnoslovenskim jezicima

Visoko-kvalitetan skup raznovrsnih naučnih publikacija

Neophodan za obučavanje kvalitetnih jezičkih modela za južnoslovenske jezike.

Ukupno 280,460 dokumenata, ukupno sa preko 4.2 milijarde reči.

Svaka JSON linija predstavlja jednu publikaciju.

Unutar svakog dokumenta su obeležene rečenice i paragrafi.

Set of South Slavic Scientific Research publications

Highly curated, High-qualityand diverse scientific publications.

Necessary for training quality language models for South Slavic languages.

A total of 280,460 documents containing over 4.2 billion words.

Each JSON line represents one publication.

All documents are paragraph and sentence-delimited.

Број докумената
Doc. count
Број реченица
Sent. count
Број речи
Word count
Удео
Share
Korpus S.T.A.R.S.
🇷🇸
23,340 27,760,829 704,000,000 16.7%
DABAR
🇭🇷
108,786 53,369,657 1,214,000,000 28.8%
Open Science Slovenia
🇸🇮 🇷🇸 🇭🇷 🇧🇦
148,334 105,099,752 2,296,000,000 54.5%
Укупно
Total
280,460 186,230,238 4,214,000,000 100%
```python from datasets import load_dataset dataset = load_dataset("procesaur/ZNANJE") ```
Editor
Mihailo Škorić
@procesaur
Citation: ```bibtex @article{skoric24korpusi, author = {\vSkori\'c, Mihailo and Jankovi\'c, Nikola}, title = {New Textual Corpora for Serbian Language Modeling}, journal = {Infotheca}, volume = {24}, issue = {1}, year = {2024}, publisher = {Zajednica biblioteka univerziteta u Srbiji, Beograd}, url = {https://arxiv.org/abs/2405.09250} } ```

Istraživanje je sprovedeno uz podršku Fonda za nauku Republike Srbije, #7276, Text Embeddings – Serbian Language Applications – TESLA.

This research was supported by the Science Fund of the Republic of Serbia, #7276, Text Embeddings - Serbian Language Applications - TESLA.