This is the second version of the Ukrainian NER corpus. You can find the first version data and documentation here
The labeled data corpus is located in the v2.0/data folder.
Total in the corpus:
- 560 texts (train: 391, test: 169)
- 21_993 NER entities
- 13 types of entities
| NashiGroshi | Bruk | Total | |
|---|---|---|---|
| ART | 319 | 316 | 635 |
| DATE | 1496 | 551 | 2047 |
| DOC | 108 | 34 | 142 |
| JOB | 1344 | 638 | 1982 |
| LOC | 1380 | 1620 | 3000 |
| MISC | 102 | 413 | 515 |
| MON | 897 | 46 | 943 |
| ORG | 4431 | 782 | 5213 |
| PCT | 186 | 77 | 263 |
| PERIOD | 341 | 255 | 596 |
| PERS | 1820 | 4415 | 6235 |
| QUANT | 276 | 106 | 382 |
| TIME | 4 | 36 | 40 |
| Total | 12704 | 9289 | 21993 |
The primary data source is the Open Corpus of Ukrainian Texts (folder bruk and the texts of the publication "Nashi Groshi" (folder ng). There are two files for each processed text from the corpus:
- a file with the extension
txtcontains the tokenized version of the text - a file with the extension
anncontains NER-annotations to this text in Brat Standoff Format (each line of the file contains 3 records separated by tabs: the annotation number, the start and end index in the text - in this case, the tokenized one - separated by a space, the entity text)
The annotation was performed by at least two annotators for each text according to the following [rules] (doc/README.md), with discrepancies in the results corrected by a third editor.
For model training and validation, we recommend using the Standard split into DEV and TEST sets.
We provide IOB-converted data using the standard breakdown. Під час цієї конвертації ми прибрали вкладені теги.
The repository also contains scripts for converting data to other formats.
This data is available for use under the terms of the "Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License"
"Корпус NER-анотацій українських текстів" by lang-uk is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
Based on a work at https://github.com/lang-uk/ner-uk.
