Datasets:
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,839 @@
|
|
| 1 |
-
---
|
| 2 |
-
|
| 3 |
-
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language:
|
| 3 |
+
- lt
|
| 4 |
+
- en
|
| 5 |
+
pretty_name: General Lithuanian Corpus
|
| 6 |
+
tags:
|
| 7 |
+
- text
|
| 8 |
+
- parquet
|
| 9 |
+
- corpus
|
| 10 |
+
- lithuanian
|
| 11 |
+
license: other
|
| 12 |
+
license_name: openrail-d
|
| 13 |
+
license_link: https://sitti.vdu.lt/newgenltu-openrail-d-license/
|
| 14 |
+
size_categories:
|
| 15 |
+
- 1M<n<10M
|
| 16 |
+
task_categories:
|
| 17 |
+
- text-generation
|
| 18 |
+
- summarization
|
| 19 |
+
configs:
|
| 20 |
+
- config_name: default
|
| 21 |
+
data_files:
|
| 22 |
+
- split: train
|
| 23 |
+
path: data/*.parquet
|
| 24 |
+
---
|
| 25 |
+
|
| 26 |
+
# Bendrasis lietuvių kalbos tekstynas / General Lithuanian Corpus
|
| 27 |
+
|
| 28 |
+
## EN [LT](#lt)
|
| 29 |
+
|
| 30 |
+
### Description
|
| 31 |
+
|
| 32 |
+
The General Lithuanian Language Corpus (hereinafter referred to as BLKT) was developed by State Digital Solutions Agency (SDSA) as part of the project “Creation of the General Lithuanian Language Corpus and Vectorized Models” (project code: 02-103-P-0001). BLKT was developed by consortium of four partners: Vytautas Magnus University (VMU), UAB “Tilde informacinės technologijos“, UAB “Neurotechnology“, and MB “Krilas“.
|
| 33 |
+
|
| 34 |
+
The project is being implemented in accordance with Progress Measure No. 05-002-01-07-08 of the 2021–2030 Lithuanian Republic State Digitalization Development Program “Developing technological solutions and tools that enable safe and convenient use of services” and meets the requirements of its activity “Development of linguistic resources for the needs of artificial intelligence technology solutions.”
|
| 35 |
+
|
| 36 |
+
### Dataset summary
|
| 37 |
+
|
| 38 |
+
- Number of Parquet files: **25**
|
| 39 |
+
- Total rows: **8,438,155**
|
| 40 |
+
- Total file size: **17.79 GB**
|
| 41 |
+
- Number of alpha words: **3,941,476,219**
|
| 42 |
+
|
| 43 |
+
### Main columns
|
| 44 |
+
|
| 45 |
+
- `alpha_word_count`: `int64`
|
| 46 |
+
- `author`: `string`
|
| 47 |
+
- `document_subtype`: `string`
|
| 48 |
+
- `document_type`: `string`
|
| 49 |
+
- `id`: `string`
|
| 50 |
+
- `language`: `string`
|
| 51 |
+
- `license`: `string`
|
| 52 |
+
- `period`: `string`
|
| 53 |
+
- `publication_date`: `string`
|
| 54 |
+
- `record_created`: `string`
|
| 55 |
+
- `source_file`: `string`
|
| 56 |
+
- `source_id`: `string`
|
| 57 |
+
- `source_name`: `string`
|
| 58 |
+
- `text`: `string`
|
| 59 |
+
- `text_char_count`: `int64`
|
| 60 |
+
- `title`: `string`
|
| 61 |
+
- `url`: `string`
|
| 62 |
+
|
| 63 |
+
### Provenance
|
| 64 |
+
|
| 65 |
+
All BLKT texts are in Lithuanian, human-written, devided into 5 text types (document_types): fiction (gro), non-fiction (neg), media (zin), speech (sak) and documents (dok). Text types are further devided into subtypes (document_subtypes) (see the table below).
|
| 66 |
+
|
| 67 |
+
| Type | Description | Subtype | Description |
|
| 68 |
+
| ----- | ----------- | ------- | -------------------------- |
|
| 69 |
+
| zin | Media | port | News portals |
|
| 70 |
+
| zin | Media | prdk | Newspapers, journals |
|
| 71 |
+
| sak | Speech | sste | Parliament transcriptions |
|
| 72 |
+
| sak | Speech | skit | Other speech transkritions |
|
| 73 |
+
| neg | Non-fiction | intt | Internet texts |
|
| 74 |
+
| neg | Non-fiction | moks | Scientific papers, books |
|
| 75 |
+
| neg | Non-fiction | kiti | Other texts |
|
| 76 |
+
| gro | Fiction | proz | Proze |
|
| 77 |
+
| gro | Fiction | poez | Poetry |
|
| 78 |
+
| dok | Documents | dkes | EU documents |
|
| 79 |
+
| dok | Documents | dklt | LR legal documents |
|
| 80 |
+
| dok | Documents | dkad | Administration documents |
|
| 81 |
+
|
| 82 |
+
The distrubution of text types:
|
| 83 |
+
|
| 84 |
+
| Text type | Word count | Part | Text count |
|
| 85 |
+
| ------------- | -------------: | ----------: | ------------: |
|
| 86 |
+
| Documents | 1509347610 | 38.29% | 1212131 |
|
| 87 |
+
| Fiction | 11767755 | 0.30% | 440 |
|
| 88 |
+
| Non-fiction | 279251467 | 7.08% | 832256 |
|
| 89 |
+
| Speech | 80483739 | 2.04% | 4563 |
|
| 90 |
+
| Media | 2060625648 | 52.28% | 6388765 |
|
| 91 |
+
| **In total** | **3941476219** | **100.00%** | **8438155** |
|
| 92 |
+
|
| 93 |
+
The texts have been collected from 36 different sources, ensuring transparency, collecting all necessary permissions and/or licenses.
|
| 94 |
+
|
| 95 |
+
| No | Source | Number of texts | Word count | Part (%) |
|
| 96 |
+
| -: | ----------------------------- | --------------: | --------------: | ----------: |
|
| 97 |
+
| 1 | 15min.lt | 559107 | 230933602 | 5.86% |
|
| 98 |
+
| 2 | autogidas.lt | 1846 | 735416 | 0.02% |
|
| 99 |
+
| 3 | bernardinai.lt | 114040 | 65085920 | 1.65% |
|
| 100 |
+
| 4 | BNS | 352946 | 83877818 | 2.13% |
|
| 101 |
+
| 5 | CRIS-DB | 6483 | 34630986 | 0.88% |
|
| 102 |
+
| 6 | delfi.lt | 1714345 | 606662831 | 15.39% |
|
| 103 |
+
| 7 | elta.lt | 1235494 | 254356849 | 6.45% |
|
| 104 |
+
| 8 | epaveldas.lt | 265 | 11339315 | 0.29% |
|
| 105 |
+
| 9 | etalpykla.vilniustech.lt.lt | 258 | 210746 | 0.01% |
|
| 106 |
+
| 10 | e-tar.lt | 307223 | 567471590 | 14.40% |
|
| 107 |
+
| 11 | eur-lex.europa.eu | 141887 | 319901553 | 8.12% |
|
| 108 |
+
| 12 | Gargždapilis | 99 | 98983 | 0.00% |
|
| 109 |
+
| 13 | gazas.lt | 5454 | 2022545 | 0.05% |
|
| 110 |
+
| 14 | gov.lt | 515173 | 162776172 | 4.13% |
|
| 111 |
+
| 15 | Hubris | 16 | 1118951 | 0.03% |
|
| 112 |
+
| 16 | Humanitarų meka | 64 | 68586 | 0.00% |
|
| 113 |
+
| 17 | Internet archive | 112 | 5592994 | 0.14% |
|
| 114 |
+
| 18 | Įv. autoriai | 182 | 1226753 | 0.03% |
|
| 115 |
+
| 19 | kaunas.kasvyksta.lt | 49662 | 15141979 | 0.38% |
|
| 116 |
+
| 20 | Lituanistika_DB | 286 | 1036134 | 0.03% |
|
| 117 |
+
| 21 | LLTI | 2 | 867891 | 0.02% |
|
| 118 |
+
| 22 | lrytas.lt | 1140602 | 380082600 | 9.64% |
|
| 119 |
+
| 23 | lrkt.lt | 1029 | 4162009 | 0.11% |
|
| 120 |
+
| 24 | lrs.lt | 134968 | 118782124 | 3.01% |
|
| 121 |
+
| 25 | lrt.lt | 761744 | 244438996 | 6.20% |
|
| 122 |
+
| 26 | MIDAS | 42 | 39747 | 0.00% |
|
| 123 |
+
| 27 | pirkimai.eviesiejipirkimai.lt | 46288 | 117844194 | 2.99% |
|
| 124 |
+
| 28 | raudonoji_knyga | 396 | 103296 | 0.00% |
|
| 125 |
+
| 29 | sena.lt | 88 | 31883 | 0.00% |
|
| 126 |
+
| 30 | Švenčionių kraštas | 11046 | 5253903 | 0.13% |
|
| 127 |
+
| 31 | tautinispaveldas.lt | 392 | 59793 | 0.00% |
|
| 128 |
+
| 32 | teismai.lt | 715704 | 499968264 | 12.68% |
|
| 129 |
+
| 33 | vdu.lt | 5937 | 709234 | 0.02% |
|
| 130 |
+
| 34 | ve.lt | 427947 | 160571897 | 4.07% |
|
| 131 |
+
| 35 | Vikipedija | 170718 | 34741743 | 0.88% |
|
| 132 |
+
| 36 | zmones.15min.lt | 16310 | 9528922 | 0.24% |
|
| 133 |
+
| | **In total** | **8438155** | **3941476219** | **100.00%** |
|
| 134 |
+
|
| 135 |
+
The texts have been published during 4 periods
|
| 136 |
+
|
| 137 |
+
| Time span | Period | Word count | Part | Text count |
|
| 138 |
+
| ----------- | ----------: | -------------: | ----------: | ----------: |
|
| 139 |
+
| 1922–1940 | 1 | 11226438 | 0.28% | 290 |
|
| 140 |
+
| 1941–1990 | 2 | 10440807 | 0.26% | 4288 |
|
| 141 |
+
| 1990–2004 | 3 | 229775041 | 5.83% | 684357 |
|
| 142 |
+
| 2008–2026 | 4 | 3690033933 | 93.62% | 7749220 |
|
| 143 |
+
| | **In total** | **3941476219** | **100.00%** | **8438155** |
|
| 144 |
+
|
| 145 |
+
|
| 146 |
+
### Licensing
|
| 147 |
+
|
| 148 |
+
This dataset is distributed under **NewGenLTU OpenRAIL-D**.
|
| 149 |
+
|
| 150 |
+
This Open RAIL-D license was specifically created by a team of researchers at Vytautas Magnus University for the publication of the projects’ results – datasets – that had been developed in the course of the research and development projects funded by the New Generation Lithuania plan. The New Generation Lithuania plan is funded by the Recovery and Resilience Facility (RRF). It was decided to establish this license in order to address certain limitations relating to the use of datasets and to ensure compliance with the EU AI Act, data protection requirements, and copyright law.
|
| 151 |
+
|
| 152 |
+
This license strives for open and responsible downstream use of the model and/or dataset(s) while ensuring that the works subject to copyright that are contained in the dataset will not be used to recreate the works but rather will be used for AI training and other technology development activities. Similarly, with the models, the license strives to ensure compliance with the EU data protection regulation thus prohibiting the recreation or extraction of personal data from the dataset.
|
| 153 |
+
|
| 154 |
+
This license governs the use of the dataset and/or the model (and their derivatives).
|
| 155 |
+
|
| 156 |
+
Please see the accompanying `LICENSE.txt` file in this repository.
|
| 157 |
+
|
| 158 |
+
### Recommended use
|
| 159 |
+
|
| 160 |
+
- Generative tasks: automatic text generation, summary generation, and the generation of creative ideas and scripts.
|
| 161 |
+
- Editing and proofreading: improving grammar, style, text clarity, and structure.
|
| 162 |
+
- Business solutions:
|
| 163 |
+
- Language editing tools.
|
| 164 |
+
- Smart search and text summarization.
|
| 165 |
+
- Creation of virtual assistants and marketing copy.
|
| 166 |
+
- Data analysis and insights into customer needs.
|
| 167 |
+
|
| 168 |
+
### Restrictions
|
| 169 |
+
|
| 170 |
+
Under the **NewGenLTU OpenRAIL-D** license, users must not use the dataset for the following purposes: discrimination; the development of weapons or military applications; automated decision making; generation of disinformation; privacy infringement; the provision of medical or health advice; the creation of malware; harassment; dishonest research; or the collection of personal data.
|
| 171 |
+
For further details, please see the accompanying `LICENSE.txt` file in this repository.
|
| 172 |
+
|
| 173 |
+
### Limitations
|
| 174 |
+
|
| 175 |
+
The developers tried their best to clean the dataset from noise, OCR errors, duplicates. However:
|
| 176 |
+
- Number of spellcheking errors:
|
| 177 |
+
- Number of foreign words:
|
| 178 |
+
- Texts from periods 1-2 (0.54%) may contain old unused words, phrases, or expressions
|
| 179 |
+
- The corpus is dominated by texts from news portals (52%) and documents (38%).
|
| 180 |
+
|
| 181 |
+
### Citation
|
| 182 |
+
|
| 183 |
+
"General Lithuanian corpus".
|
| 184 |
+
|
| 185 |
+
## Shared technical information / Bendra techninė informacija
|
| 186 |
+
|
| 187 |
+
### Schema
|
| 188 |
+
|
| 189 |
+
| Column | Main type | Present in files | Null count |
|
| 190 |
+
|---|---|---|---|
|
| 191 |
+
| alpha_word_count | int64 | 25/25 | 0 |
|
| 192 |
+
| author | null | 25/25 | 8381572 |
|
| 193 |
+
| document_subtype | string | 25/25 | 0 |
|
| 194 |
+
| document_type | string | 25/25 | 0 |
|
| 195 |
+
| id | string | 25/25 | 0 |
|
| 196 |
+
| language | string | 25/25 | 0 |
|
| 197 |
+
| license | string | 25/25 | 0 |
|
| 198 |
+
| period | string | 25/25 | 0 |
|
| 199 |
+
| publication_date | string | 25/25 | 509009 |
|
| 200 |
+
| record_created | string | 25/25 | 0 |
|
| 201 |
+
| source_file | string | 25/25 | 0 |
|
| 202 |
+
| source_id | string | 25/25 | 0 |
|
| 203 |
+
| source_name | string | 25/25 | 0 |
|
| 204 |
+
| text | string | 25/25 | 0 |
|
| 205 |
+
| text_char_count | int64 | 25/25 | 0 |
|
| 206 |
+
| title | string | 25/25 | 8093921 |
|
| 207 |
+
| url | string | 25/25 | 311723 |
|
| 208 |
+
|
| 209 |
+
### Files
|
| 210 |
+
|
| 211 |
+
| File | Rows | Size | Row groups |
|
| 212 |
+
|---|---:|---:|---:|
|
| 213 |
+
| BLKT_corpus-00000.parquet | 255000 | 486.48 MB | 1 |
|
| 214 |
+
| BLKT_corpus-00001.parquet | 525000 | 469.60 MB | 1 |
|
| 215 |
+
| BLKT_corpus-00002.parquet | 235000 | 403.29 MB | 1 |
|
| 216 |
+
| BLKT_corpus-00003.parquet | 235000 | 403.66 MB | 1 |
|
| 217 |
+
| BLKT_corpus-00004.parquet | 175000 | 370.09 MB | 1 |
|
| 218 |
+
| BLKT_corpus-00005.parquet | 55000 | 376.28 MB | 1 |
|
| 219 |
+
| BLKT_corpus-00006.parquet | 80000 | 355.37 MB | 1 |
|
| 220 |
+
| BLKT_corpus-00007.parquet | 80000 | 372.77 MB | 1 |
|
| 221 |
+
| BLKT_corpus-00008.parquet | 90000 | 380.82 MB | 1 |
|
| 222 |
+
| BLKT_corpus-00009.parquet | 85000 | 383.33 MB | 1 |
|
| 223 |
+
| BLKT_corpus-00010.parquet | 215000 | 381.49 MB | 1 |
|
| 224 |
+
| BLKT_corpus-00011.parquet | 405000 | 493.28 MB | 1 |
|
| 225 |
+
| BLKT_corpus-00012.parquet | 330000 | 471.21 MB | 1 |
|
| 226 |
+
| BLKT_corpus-00013.parquet | 395000 | 592.07 MB | 1 |
|
| 227 |
+
| BLKT_corpus-00014.parquet | 345000 | 581.43 MB | 1 |
|
| 228 |
+
| BLKT_corpus-00015.parquet | 510000 | 585.07 MB | 1 |
|
| 229 |
+
| BLKT_corpus-00016.parquet | 500000 | 588.28 MB | 1 |
|
| 230 |
+
| BLKT_corpus-00017.parquet | 480000 | 582.85 MB | 1 |
|
| 231 |
+
| BLKT_corpus-00018.parquet | 425000 | 576.71 MB | 1 |
|
| 232 |
+
| BLKT_corpus-00019.parquet | 530000 | 542.38 MB | 1 |
|
| 233 |
+
| BLKT_corpus-00020.parquet | 485000 | 560.00 MB | 1 |
|
| 234 |
+
| BLKT_corpus-00021.parquet | 370000 | 590.74 MB | 1 |
|
| 235 |
+
| BLKT_corpus-00022.parquet | 455000 | 553.52 MB | 1 |
|
| 236 |
+
| BLKT_corpus-00023.parquet | 710000 | 499.35 MB | 1 |
|
| 237 |
+
| BLKT_corpus-00024.parquet | 468155 | 388.07 MB | 1 |
|
| 238 |
+
|
| 239 |
+
### Loading the dataset / Duomenų rinkinio įkėlimas
|
| 240 |
+
|
| 241 |
+
Example using the `datasets` library:
|
| 242 |
+
|
| 243 |
+
```python
|
| 244 |
+
from datasets import load_dataset
|
| 245 |
+
|
| 246 |
+
ds = load_dataset("parquet", data_files="hf://datasets/VSSA-SDSA/lithuanian-general-corpus/data/*.parquet")
|
| 247 |
+
print(ds)
|
| 248 |
+
```
|
| 249 |
+
|
| 250 |
+
Example using pandas:
|
| 251 |
+
|
| 252 |
+
```python
|
| 253 |
+
import pandas as pd
|
| 254 |
+
|
| 255 |
+
# Example for one file
|
| 256 |
+
df = pd.read_parquet("data/your_file.parquet")
|
| 257 |
+
print(df.head())
|
| 258 |
+
```
|
| 259 |
+
|
| 260 |
+
### Files with metadata read errors / Failai su metaduomenų nuskaitymo klaidomis
|
| 261 |
+
|
| 262 |
+
None.
|
| 263 |
+
<a id=lt></a>
|
| 264 |
+
## LT
|
| 265 |
+
|
| 266 |
+
### Aprašas
|
| 267 |
+
|
| 268 |
+
Bendrasis lietuvių kalbos tekstynas (toliau – BLKT) sukurtas Valstybės skaitmeninių sprendimų agentūrai (VSSA) įgyvendinant projektą „Bendrojo lietuvių kalbos tekstyno ir vektorizuotų modelių sukūrimas“ (projekto kodas: 02-103-P-0001). BLKT sukūrė keturių partnerių konsorciumas: Vytauto Didžiojo universitetas (VDU), UAB „Neurotechnology“, UAB „Tilde informacinės technologijos“ ir MB „Krilas“.
|
| 269 |
+
|
| 270 |
+
Projektas vykdomas vadovaujantis 2021–2030 metų Lietuvos Respublikos valstybės skaitmeninimo plėtros programos pažangos priemone Nr. 05-002-01-07-08 „Kurti technologinius sprendimus ir įrankius, leidžiančius saugiai ir patogiai naudotis paslaugomis“ ir atliepia jos veiklos „Kalbinių išteklių dirbtinio intelekto technologijų sprendimų poreikiams plėtra“ reikalavimus.
|
| 271 |
+
|
| 272 |
+
### Duomenų rinkinio santrauka
|
| 273 |
+
|
| 274 |
+
- Žodžių skaičius: **3 941 476 219**
|
| 275 |
+
- Bendras eilučių (tekstų) skaičius: **8 438 155**
|
| 276 |
+
- Parquet failų skaičius: **25**
|
| 277 |
+
- Bendras failų dydis: **17,79 GB**
|
| 278 |
+
|
| 279 |
+
### Pagrindiniai stulpeliai
|
| 280 |
+
|
| 281 |
+
- `alpha_word_count`: `int64`
|
| 282 |
+
- `author`: `string`
|
| 283 |
+
- `document_subtype`: `string`
|
| 284 |
+
- `document_type`: `string`
|
| 285 |
+
- `id`: `string`
|
| 286 |
+
- `language`: `string`
|
| 287 |
+
- `license`: `string`
|
| 288 |
+
- `period`: `string`
|
| 289 |
+
- `publication_date`: `string`
|
| 290 |
+
- `record_created`: `string`
|
| 291 |
+
- `source_file`: `string`
|
| 292 |
+
- `source_id`: `string`
|
| 293 |
+
- `source_name`: `string`
|
| 294 |
+
- `text`: `string`
|
| 295 |
+
- `text_char_count`: `string`
|
| 296 |
+
- `title`: `null`
|
| 297 |
+
- `url`: `string`
|
| 298 |
+
|
| 299 |
+
### Kilmė ir sudarymas
|
| 300 |
+
|
| 301 |
+
Visi BLKT tekstai skirstomi į 5 teksto tipus (document_types): grožinius (gro), negrožinius (neg), žiniasklaidos (zin), sakytinės kalbos (sak) ir dokumentus (dok). Teksto tipai toliau skirstomi į potipius (subtypes) (žr. lentelę).
|
| 302 |
+
|
| 303 |
+
| Tipas | Apibūdinimas | Potipis | Apibūdinimas |
|
| 304 |
+
| ----- | ------------------------ | ------- | ------------------------------------- |
|
| 305 |
+
| zin | Žiniasklaidos tekstai | port | Interneto naujienų portalai |
|
| 306 |
+
| zin | Žiniasklaidos tekstai | prdk | Laikraščiai, žurnalai |
|
| 307 |
+
| sak | Sakytinės kalbos tekstai | sste | Seimo stenogramos |
|
| 308 |
+
| sak | Sakytinės kalbos tekstai | skit | Kitos sakytinės kalbos transkripcijos |
|
| 309 |
+
| neg | Negrožiniai tekstai | intt | Interneto tekstai |
|
| 310 |
+
| neg | Negrožiniai tekstai | moks | Mokslo straipsniai, knygos |
|
| 311 |
+
| neg | Negrožiniai tekstai | kiti | Kiti tekstai |
|
| 312 |
+
| gro | Grožiniai tekstai | proz | Proza |
|
| 313 |
+
| gro | Grožiniai tekstai | poez | Poezija |
|
| 314 |
+
| dok | Dokumentai | dkes | ES dokumentai |
|
| 315 |
+
| dok | Dokumentai | dklt | LR teisiniai dokumentai |
|
| 316 |
+
| dok | Dokumentai | dkad | Administraciniai dokumentai |
|
| 317 |
+
|
| 318 |
+
Teksto tipų pasiskirstymas:
|
| 319 |
+
|
| 320 |
+
| Teksto tipas | Žodžių sk. | Dalis | Tekstų sk. |
|
| 321 |
+
| --------------------- | -------------: | ----------: | ----------: |
|
| 322 |
+
| Dokumentai | 1509347610 | 38.29% | 1212131 |
|
| 323 |
+
| Grožiniai tekstai | 11767755 | 0.30% | 440 |
|
| 324 |
+
| Negrožiniai tekstai | 279251467 | 7.08% | 832256 |
|
| 325 |
+
| Sakytinės k. tekstai | 80483739 | 2.04% | 4563 |
|
| 326 |
+
| Žiniasklaidos tekstai | 2060625648 | 52.28% | 6388765 |
|
| 327 |
+
| **Iš viso** | **3941476219** | **100.00%** | **8438155** |
|
| 328 |
+
|
| 329 |
+
Tekstai surinkti iš 36 skirtingų šaltinių užtikrinant skaidrumą, iš duomenų tiekėjų gavus visus reikalingus sutikimus ir licencijas.
|
| 330 |
+
|
| 331 |
+
| Nr | Šaltinis | Tekstų skaičius | Žodžių skaičius | Dalis (%) |
|
| 332 |
+
| -: | ----------------------------- | --------------: | --------------: | ----------: |
|
| 333 |
+
| 1 | 15min.lt | 559107 | 230933602 | 5,86% |
|
| 334 |
+
| 2 | autogidas.lt | 1846 | 735416 | 0,02% |
|
| 335 |
+
| 3 | bernardinai.lt | 114040 | 65085920 | 1,65% |
|
| 336 |
+
| 4 | BNS | 352946 | 83877818 | 2,13% |
|
| 337 |
+
| 5 | CRIS-DB | 6483 | 34630986 | 0,88% |
|
| 338 |
+
| 6 | delfi.lt | 1714345 | 606662831 | 15,39% |
|
| 339 |
+
| 7 | elta.lt | 1235494 | 254356849 | 6,45% |
|
| 340 |
+
| 8 | epaveldas.lt | 265 | 11339315 | 0,29% |
|
| 341 |
+
| 9 | etalpykla.vilniustech.lt.lt | 258 | 210746 | 0,01% |
|
| 342 |
+
| 10 | e-tar.lt | 307223 | 567471590 | 14,40% |
|
| 343 |
+
| 11 | eur-lex.europa.eu | 141887 | 319901553 | 8,12% |
|
| 344 |
+
| 12 | Gargždapilis | 99 | 98983 | 0,00% |
|
| 345 |
+
| 13 | gazas.lt | 5454 | 2022545 | 0,05% |
|
| 346 |
+
| 14 | gov.lt | 515173 | 162776172 | 4,13% |
|
| 347 |
+
| 15 | Hubris | 16 | 1118951 | 0,03% |
|
| 348 |
+
| 16 | Humanitarų meka | 64 | 68586 | 0,00% |
|
| 349 |
+
| 17 | Internet archive | 112 | 5592994 | 0,14% |
|
| 350 |
+
| 18 | Įv. autoriai | 182 | 1226753 | 0,03% |
|
| 351 |
+
| 19 | kaunas.kasvyksta.lt | 49662 | 15141979 | 0,38% |
|
| 352 |
+
| 20 | Lituanistika_DB | 286 | 1036134 | 0,03% |
|
| 353 |
+
| 21 | LLTI | 2 | 867891 | 0,02% |
|
| 354 |
+
| 22 | lrytas.lt | 1140602 | 380082600 | 9,64% |
|
| 355 |
+
| 23 | lrkt.lt | 1029 | 4162009 | 0,11% |
|
| 356 |
+
| 24 | lrs.lt | 134968 | 118782124 | 3,01% |
|
| 357 |
+
| 25 | lrt.lt | 761744 | 244438996 | 6,20% |
|
| 358 |
+
| 26 | MIDAS | 42 | 39747 | 0,00% |
|
| 359 |
+
| 27 | pirkimai.eviesiejipirkimai.lt | 46288 | 117844194 | 2,99% |
|
| 360 |
+
| 28 | raudonoji_knyga | 396 | 103296 | 0,00% |
|
| 361 |
+
| 29 | sena.lt | 88 | 31883 | 0,00% |
|
| 362 |
+
| 30 | Švenčionių kraštas | 11046 | 5253903 | 0,13% |
|
| 363 |
+
| 31 | tautinispaveldas.lt | 392 | 59793 | 0,00% |
|
| 364 |
+
| 32 | teismai.lt | 715704 | 499968264 | 12,68% |
|
| 365 |
+
| 33 | vdu.lt | 5937 | 709234 | 0,02% |
|
| 366 |
+
| 34 | ve.lt | 427947 | 160571897 | 4,07% |
|
| 367 |
+
| 35 | Vikipedija | 170718 | 34741743 | 0,88% |
|
| 368 |
+
| 36 | zmones.15min.lt | 16310 | 9528922 | 0,24% |
|
| 369 |
+
| | **Iš viso** | **8438155** | **3941476219** | **100,00%** |
|
| 370 |
+
|
| 371 |
+
Surinkti tekstai publikuoti
|
| 372 |
+
|
| 373 |
+
| Laikotarpis | Žyma | Žodžių sk. | Dalis | Tekstų sk. |
|
| 374 |
+
| ----------- | ----------: | -------------: | ----------: | ----------: |
|
| 375 |
+
| 1922–1940 | 1 | 11226438 | 0,28% | 290 |
|
| 376 |
+
| 1941–1990 | 2 | 10440807 | 0,26% | 4288 |
|
| 377 |
+
| 1990–2004 | 3 | 229775041 | 5,83% | 684357 |
|
| 378 |
+
| 2008–2026 | 4 | 3690033933 | 93,62% | 7749220 |
|
| 379 |
+
| | **Iš viso** | **3941476219** | **100,00%** | **8438155** |
|
| 380 |
+
|
| 381 |
+
|
| 382 |
+
### Licencija
|
| 383 |
+
|
| 384 |
+
Šis duomenų rinkinys yra platinamas pagal **NewGenLTU OpenRAIL-D** licenciją.
|
| 385 |
+
|
| 386 |
+
Ši „Open RAIL-D“ licencija buvo specialiai sukurta Vytauto Didžiojo universiteto tyrėjų komandos, siekiant skelbti projekto rezultatus – duomenų rinkinius, – kurie buvo parengti vykdant mokslinių tyrimų ir plėtros projektus, finansuojamus pagal programą „Naujos kartos Lietuvoje“. Programa „Naujos kartos Lietuva“ finansuojama iš Ekonomikos gaivinimo ir atsparumo didinimo priemonės (RRF). Buvo nuspręsta sukurti šią licenciją siekiant pašalinti tam tikrus duomenų rinkinių naudojimo apribojimus ir užtikrinti atitiktį ES AI aktui, duomenų apsaugos reikalavimams bei autorių teisių įstatymams.
|
| 387 |
+
|
| 388 |
+
Šia licencija siekiama atviro ir atsakingo modelio ir (arba) duomenų rinkinių tolesnio naudojimo, tuo pačiu užtikrinant, kad duomenų rinkinyje esantys kūriniai, kuriems taikomos autorių teisės, nebūtų naudojami kūriniams atkurti, o būtų naudojami AI mokymui ir kitai technologijų plėtros veiklai. Panašiai, kaip ir modelių atveju, licencija siekia užtikrinti atitiktį ES duomenų apsaugos reglamentui, taigi draudžia asmens duomenų atkūrimą ar išgavimą iš duomenų rinkinio.
|
| 389 |
+
|
| 390 |
+
Ši licencija reglamentuoja duomenų rinkinio ir (arba) modelio (ir jų išvestinių kūrinių) naudojimą.
|
| 391 |
+
Prašome peržiūrėti pridedamą failą „LICENSE.txt“ šiame saugykloje.
|
| 392 |
+
|
| 393 |
+
### Rekomenduojamas naudojimas
|
| 394 |
+
|
| 395 |
+
- Generatyvinė užduotis: automatinį tekstų kūrimas, apibendrinimų generavimas, kūrybinių idėjų ir scenarijų generavimas.
|
| 396 |
+
- Redagavimo užduotis ir korektūra: gramatikos, stiliaus teksto aiškumo ir struktūros tobulinimas.
|
| 397 |
+
- Verslo sprendimai:
|
| 398 |
+
- Kalbos redagavimo įrankiai
|
| 399 |
+
- Išmanioji paieška ir teksto santraukų generavimas.
|
| 400 |
+
- Virtualių asistentų, rinkodaros tekstų kūrimas.
|
| 401 |
+
- Duomenų analizė ir klientų poreikių įžvalgos.
|
| 402 |
+
|
| 403 |
+
### Apribojimai
|
| 404 |
+
|
| 405 |
+
Pagal **NewGenLTU OpenRAIL-D** licenciją vartotojai negali naudoti šio duomenų rinkinio šiais tikslais: diskriminacijai; ginklų ar karinių programų kūrimui; automatizuotam sprendimų priėmimui; dezinformacijos skleidimui; privatumo pažeidimams; medicininių ar sveikatos patarimų teikimui; kenkėjiškų programų kūrimui; priekabiavimui; nesąžiningiems tyrimams; arba asmens duomenų rinkimui.
|
| 406 |
+
Daugiau informacijos rasite šioje saugykloje esančiame pridedamame faile `LICENSE.txt`.
|
| 407 |
+
|
| 408 |
+
### Ribotumai
|
| 409 |
+
|
| 410 |
+
Kūrėjai dėjo visas pastangas, kad tekstyną išvalytų nuo šiukšlių, klaidų ir dubliažų. Tačiau:
|
| 411 |
+
- Rašybos klaidų dalis: 0,46%
|
| 412 |
+
- Kitų kalbų žodžių dalis: 0,07%
|
| 413 |
+
- 1–2 laikotarpių tekstuose (0,54 %) gali būti senų, nebevartojamų žodžių, frazių ar posakių
|
| 414 |
+
- Tekstye vyrauja tekstai iš naujienų portalų (52 %) ir dokumentų (38 %).
|
| 415 |
+
|
| 416 |
+
### Citavimas
|
| 417 |
+
|
| 418 |
+
"Bendrasis lietuvių kalbos tekstynas".
|
| 419 |
+
|
| 420 |
+
---
|
| 421 |
+
language:
|
| 422 |
+
- lt
|
| 423 |
+
- en
|
| 424 |
+
pretty_name: General Lithuanian Corpus
|
| 425 |
+
tags:
|
| 426 |
+
- text
|
| 427 |
+
- parquet
|
| 428 |
+
- corpus
|
| 429 |
+
- lithuanian
|
| 430 |
+
license: other
|
| 431 |
+
license_name: NewGenLTU OpenRAIL-D
|
| 432 |
+
license_link: http://sitti.vdu.lt/newgenltu-openrail-d-license/
|
| 433 |
+
size_categories:
|
| 434 |
+
- 1M<n<10M
|
| 435 |
+
task_categories:
|
| 436 |
+
- text-generation
|
| 437 |
+
- language-modeling
|
| 438 |
+
configs:
|
| 439 |
+
- config_name: default
|
| 440 |
+
data_files:
|
| 441 |
+
- split: train
|
| 442 |
+
path: data/*.parquet
|
| 443 |
+
---
|
| 444 |
+
|
| 445 |
+
# Bendrasis lietuvių kalbos tekstynas / General Lithuanian Corpus
|
| 446 |
+
|
| 447 |
+
## EN [LT](#lt)
|
| 448 |
+
|
| 449 |
+
### Description
|
| 450 |
+
|
| 451 |
+
The General Lithuanian Language Corpus (hereinafter referred to as BLKT) was developed by State Digital Solutions Agency (SDSA) as part of the project “Creation of the General Lithuanian Language Corpus and Vectorized Models” (project code: 02-103-P-0001). BLKT was developed by consortium of four partners: Vytautas Magnus University (VMU), UAB “Tilde informacinės technologijos“, UAB “Neurotechnology“, and MB “Krilas“.
|
| 452 |
+
|
| 453 |
+
The project is being implemented in accordance with Progress Measure No. 05-002-01-07-08 of the 2021–2030 Lithuanian Republic State Digitalization Development Program “Developing technological solutions and tools that enable safe and convenient use of services” and meets the requirements of its activity “Development of linguistic resources for the needs of artificial intelligence technology solutions.”
|
| 454 |
+
|
| 455 |
+
### Dataset summary
|
| 456 |
+
|
| 457 |
+
- Number of Parquet files: **25**
|
| 458 |
+
- Total rows: **8,438,155**
|
| 459 |
+
- Total file size: **17.79 GB**
|
| 460 |
+
- Number of alpha words: **3,941,476,219**
|
| 461 |
+
|
| 462 |
+
### Main columns
|
| 463 |
+
|
| 464 |
+
- `alpha_word_count`: `int64`
|
| 465 |
+
- `author`: `string`
|
| 466 |
+
- `document_subtype`: `string`
|
| 467 |
+
- `document_type`: `string`
|
| 468 |
+
- `id`: `string`
|
| 469 |
+
- `language`: `string`
|
| 470 |
+
- `license`: `string`
|
| 471 |
+
- `period`: `string`
|
| 472 |
+
- `publication_date`: `string`
|
| 473 |
+
- `record_created`: `string`
|
| 474 |
+
- `source_file`: `string`
|
| 475 |
+
- `source_id`: `string`
|
| 476 |
+
- `source_name`: `string`
|
| 477 |
+
- `text`: `string`
|
| 478 |
+
- `text_char_count`: `int64`
|
| 479 |
+
- `title`: `string`
|
| 480 |
+
- `url`: `string`
|
| 481 |
+
|
| 482 |
+
### Provenance
|
| 483 |
+
|
| 484 |
+
All BLKT texts are in Lithuanian, human-written, devided into 5 text types (document_types): fiction (gro), non-fiction (neg), media (zin), speech (sak) and documents (dok). Text types are further devided into subtypes (document_subtypes) (see the table below).
|
| 485 |
+
|
| 486 |
+
| Type | Description | Subtype | Description |
|
| 487 |
+
| ----- | ----------- | ------- | -------------------------- |
|
| 488 |
+
| zin | Media | port | News portals |
|
| 489 |
+
| zin | Media | prdk | Newspapers, journals |
|
| 490 |
+
| sak | Speech | sste | Parliament transcriptions |
|
| 491 |
+
| sak | Speech | skit | Other speech transkritions |
|
| 492 |
+
| neg | Non-fiction | intt | Internet texts |
|
| 493 |
+
| neg | Non-fiction | moks | Scientific papers, books |
|
| 494 |
+
| neg | Non-fiction | kiti | Other texts |
|
| 495 |
+
| gro | Fiction | proz | Proze |
|
| 496 |
+
| gro | Fiction | poez | Poetry |
|
| 497 |
+
| dok | Documents | dkes | EU documents |
|
| 498 |
+
| dok | Documents | dklt | LR legal documents |
|
| 499 |
+
| dok | Documents | dkad | Administration documents |
|
| 500 |
+
|
| 501 |
+
The distrubution of text types:
|
| 502 |
+
|
| 503 |
+
| Text type | Word count | Part | Text count |
|
| 504 |
+
| ------------- | -------------: | ----------: | ------------: |
|
| 505 |
+
| Documents | 1509347610 | 38.29% | 1212131 |
|
| 506 |
+
| Fiction | 11767755 | 0.30% | 440 |
|
| 507 |
+
| Non-fiction | 279251467 | 7.08% | 832256 |
|
| 508 |
+
| Speech | 80483739 | 2.04% | 4563 |
|
| 509 |
+
| Media | 2060625648 | 52.28% | 6388765 |
|
| 510 |
+
| **In total** | **3941476219** | **100.00%** | **8438155** |
|
| 511 |
+
|
| 512 |
+
The texts have been collected from 36 different sources, ensuring transparency, collecting all necessary permissions and/or licenses.
|
| 513 |
+
|
| 514 |
+
| No | Source | Number of texts | Word count | Part (%) |
|
| 515 |
+
| -: | ----------------------------- | --------------: | --------------: | ----------: |
|
| 516 |
+
| 1 | 15min.lt | 559107 | 230933602 | 5.86% |
|
| 517 |
+
| 2 | autogidas.lt | 1846 | 735416 | 0.02% |
|
| 518 |
+
| 3 | bernardinai.lt | 114040 | 65085920 | 1.65% |
|
| 519 |
+
| 4 | BNS | 352946 | 83877818 | 2.13% |
|
| 520 |
+
| 5 | CRIS-DB | 6483 | 34630986 | 0.88% |
|
| 521 |
+
| 6 | delfi.lt | 1714345 | 606662831 | 15.39% |
|
| 522 |
+
| 7 | elta.lt | 1235494 | 254356849 | 6.45% |
|
| 523 |
+
| 8 | epaveldas.lt | 265 | 11339315 | 0.29% |
|
| 524 |
+
| 9 | etalpykla.vilniustech.lt.lt | 258 | 210746 | 0.01% |
|
| 525 |
+
| 10 | e-tar.lt | 307223 | 567471590 | 14.40% |
|
| 526 |
+
| 11 | eur-lex.europa.eu | 141887 | 319901553 | 8.12% |
|
| 527 |
+
| 12 | Gargždapilis | 99 | 98983 | 0.00% |
|
| 528 |
+
| 13 | gazas.lt | 5454 | 2022545 | 0.05% |
|
| 529 |
+
| 14 | gov.lt | 515173 | 162776172 | 4.13% |
|
| 530 |
+
| 15 | Hubris | 16 | 1118951 | 0.03% |
|
| 531 |
+
| 16 | Humanitarų meka | 64 | 68586 | 0.00% |
|
| 532 |
+
| 17 | Internet archive | 112 | 5592994 | 0.14% |
|
| 533 |
+
| 18 | Įv. autoriai | 182 | 1226753 | 0.03% |
|
| 534 |
+
| 19 | kaunas.kasvyksta.lt | 49662 | 15141979 | 0.38% |
|
| 535 |
+
| 20 | Lituanistika_DB | 286 | 1036134 | 0.03% |
|
| 536 |
+
| 21 | LLTI | 2 | 867891 | 0.02% |
|
| 537 |
+
| 22 | lrytas.lt | 1140602 | 380082600 | 9.64% |
|
| 538 |
+
| 23 | lrkt.lt | 1029 | 4162009 | 0.11% |
|
| 539 |
+
| 24 | lrs.lt | 134968 | 118782124 | 3.01% |
|
| 540 |
+
| 25 | lrt.lt | 761744 | 244438996 | 6.20% |
|
| 541 |
+
| 26 | MIDAS | 42 | 39747 | 0.00% |
|
| 542 |
+
| 27 | pirkimai.eviesiejipirkimai.lt | 46288 | 117844194 | 2.99% |
|
| 543 |
+
| 28 | raudonoji_knyga | 396 | 103296 | 0.00% |
|
| 544 |
+
| 29 | sena.lt | 88 | 31883 | 0.00% |
|
| 545 |
+
| 30 | Švenčionių kraštas | 11046 | 5253903 | 0.13% |
|
| 546 |
+
| 31 | tautinispaveldas.lt | 392 | 59793 | 0.00% |
|
| 547 |
+
| 32 | teismai.lt | 715704 | 499968264 | 12.68% |
|
| 548 |
+
| 33 | vdu.lt | 5937 | 709234 | 0.02% |
|
| 549 |
+
| 34 | ve.lt | 427947 | 160571897 | 4.07% |
|
| 550 |
+
| 35 | Vikipedija | 170718 | 34741743 | 0.88% |
|
| 551 |
+
| 36 | zmones.15min.lt | 16310 | 9528922 | 0.24% |
|
| 552 |
+
| | **In total** | **8438155** | **3941476219** | **100.00%** |
|
| 553 |
+
|
| 554 |
+
The texts have been published during 4 periods
|
| 555 |
+
|
| 556 |
+
| Time span | Period | Word count | Part | Text count |
|
| 557 |
+
| ----------- | ----------: | -------------: | ----------: | ----------: |
|
| 558 |
+
| 1922–1940 | 1 | 11226438 | 0.28% | 290 |
|
| 559 |
+
| 1941–1990 | 2 | 10440807 | 0.26% | 4288 |
|
| 560 |
+
| 1990–2004 | 3 | 229775041 | 5.83% | 684357 |
|
| 561 |
+
| 2008–2026 | 4 | 3690033933 | 93.62% | 7749220 |
|
| 562 |
+
| | **In total** | **3941476219** | **100.00%** | **8438155** |
|
| 563 |
+
|
| 564 |
+
|
| 565 |
+
### Licensing
|
| 566 |
+
|
| 567 |
+
This dataset is distributed under **NewGenLTU OpenRAIL-D**.
|
| 568 |
+
|
| 569 |
+
This Open RAIL-D license was specifically created by a team of researchers at Vytautas Magnus University for the publication of the projects’ results – datasets – that had been developed in the course of the research and development projects funded by the New Generation Lithuania plan. The New Generation Lithuania plan is funded by the Recovery and Resilience Facility (RRF). It was decided to establish this license in order to address certain limitations relating to the use of datasets and to ensure compliance with the EU AI Act, data protection requirements, and copyright law.
|
| 570 |
+
|
| 571 |
+
This license strives for open and responsible downstream use of the model and/or dataset(s) while ensuring that the works subject to copyright that are contained in the dataset will not be used to recreate the works but rather will be used for AI training and other technology development activities. Similarly, with the models, the license strives to ensure compliance with the EU data protection regulation thus prohibiting the recreation or extraction of personal data from the dataset.
|
| 572 |
+
|
| 573 |
+
This license governs the use of the dataset and/or the model (and their derivatives).
|
| 574 |
+
|
| 575 |
+
Please see the accompanying `LICENSE.txt` file in this repository.
|
| 576 |
+
|
| 577 |
+
### Recommended use
|
| 578 |
+
|
| 579 |
+
- Generative tasks: automatic text generation, summary generation, and the generation of creative ideas and scripts.
|
| 580 |
+
- Editing and proofreading: improving grammar, style, text clarity, and structure.
|
| 581 |
+
- Business solutions:
|
| 582 |
+
- Language editing tools.
|
| 583 |
+
- Smart search and text summarization.
|
| 584 |
+
- Creation of virtual assistants and marketing copy.
|
| 585 |
+
- Data analysis and insights into customer needs.
|
| 586 |
+
|
| 587 |
+
### Restrictions
|
| 588 |
+
|
| 589 |
+
Under the **NewGenLTU OpenRAIL-D** license, users must not use the dataset for the following purposes: discrimination; the development of weapons or military applications; automated decision making; generation of disinformation; privacy infringement; the provision of medical or health advice; the creation of malware; harassment; dishonest research; or the collection of personal data.
|
| 590 |
+
For further details, please see the accompanying `LICENSE.txt` file in this repository.
|
| 591 |
+
|
| 592 |
+
### Limitations
|
| 593 |
+
|
| 594 |
+
The developers tried their best to clean the dataset from noise, OCR errors, duplicates. However:
|
| 595 |
+
- Number of spellcheking errors:
|
| 596 |
+
- Number of foreign words:
|
| 597 |
+
- Texts from periods 1-2 (0.54%) may contain old unused words, phrases, or expressions
|
| 598 |
+
- The corpus is dominated by texts from news portals (52%) and documents (38%).
|
| 599 |
+
|
| 600 |
+
### Citation
|
| 601 |
+
|
| 602 |
+
"General Lithuanian corpus".
|
| 603 |
+
|
| 604 |
+
## Shared technical information / Bendra techninė informacija
|
| 605 |
+
|
| 606 |
+
### Schema
|
| 607 |
+
|
| 608 |
+
| Column | Main type | Present in files | Null count |
|
| 609 |
+
|---|---|---|---|
|
| 610 |
+
| alpha_word_count | int64 | 25/25 | 0 |
|
| 611 |
+
| author | null | 25/25 | 8381572 |
|
| 612 |
+
| document_subtype | string | 25/25 | 0 |
|
| 613 |
+
| document_type | string | 25/25 | 0 |
|
| 614 |
+
| id | string | 25/25 | 0 |
|
| 615 |
+
| language | string | 25/25 | 0 |
|
| 616 |
+
| license | string | 25/25 | 0 |
|
| 617 |
+
| period | string | 25/25 | 0 |
|
| 618 |
+
| publication_date | string | 25/25 | 509009 |
|
| 619 |
+
| record_created | string | 25/25 | 0 |
|
| 620 |
+
| source_file | string | 25/25 | 0 |
|
| 621 |
+
| source_id | string | 25/25 | 0 |
|
| 622 |
+
| source_name | string | 25/25 | 0 |
|
| 623 |
+
| text | string | 25/25 | 0 |
|
| 624 |
+
| text_char_count | int64 | 25/25 | 0 |
|
| 625 |
+
| title | string | 25/25 | 8093921 |
|
| 626 |
+
| url | string | 25/25 | 311723 |
|
| 627 |
+
|
| 628 |
+
### Files
|
| 629 |
+
|
| 630 |
+
| File | Rows | Size | Row groups |
|
| 631 |
+
|---|---:|---:|---:|
|
| 632 |
+
| BLKT_corpus-00000.parquet | 255000 | 486.48 MB | 1 |
|
| 633 |
+
| BLKT_corpus-00001.parquet | 525000 | 469.60 MB | 1 |
|
| 634 |
+
| BLKT_corpus-00002.parquet | 235000 | 403.29 MB | 1 |
|
| 635 |
+
| BLKT_corpus-00003.parquet | 235000 | 403.66 MB | 1 |
|
| 636 |
+
| BLKT_corpus-00004.parquet | 175000 | 370.09 MB | 1 |
|
| 637 |
+
| BLKT_corpus-00005.parquet | 55000 | 376.28 MB | 1 |
|
| 638 |
+
| BLKT_corpus-00006.parquet | 80000 | 355.37 MB | 1 |
|
| 639 |
+
| BLKT_corpus-00007.parquet | 80000 | 372.77 MB | 1 |
|
| 640 |
+
| BLKT_corpus-00008.parquet | 90000 | 380.82 MB | 1 |
|
| 641 |
+
| BLKT_corpus-00009.parquet | 85000 | 383.33 MB | 1 |
|
| 642 |
+
| BLKT_corpus-00010.parquet | 215000 | 381.49 MB | 1 |
|
| 643 |
+
| BLKT_corpus-00011.parquet | 405000 | 493.28 MB | 1 |
|
| 644 |
+
| BLKT_corpus-00012.parquet | 330000 | 471.21 MB | 1 |
|
| 645 |
+
| BLKT_corpus-00013.parquet | 395000 | 592.07 MB | 1 |
|
| 646 |
+
| BLKT_corpus-00014.parquet | 345000 | 581.43 MB | 1 |
|
| 647 |
+
| BLKT_corpus-00015.parquet | 510000 | 585.07 MB | 1 |
|
| 648 |
+
| BLKT_corpus-00016.parquet | 500000 | 588.28 MB | 1 |
|
| 649 |
+
| BLKT_corpus-00017.parquet | 480000 | 582.85 MB | 1 |
|
| 650 |
+
| BLKT_corpus-00018.parquet | 425000 | 576.71 MB | 1 |
|
| 651 |
+
| BLKT_corpus-00019.parquet | 530000 | 542.38 MB | 1 |
|
| 652 |
+
| BLKT_corpus-00020.parquet | 485000 | 560.00 MB | 1 |
|
| 653 |
+
| BLKT_corpus-00021.parquet | 370000 | 590.74 MB | 1 |
|
| 654 |
+
| BLKT_corpus-00022.parquet | 455000 | 553.52 MB | 1 |
|
| 655 |
+
| BLKT_corpus-00023.parquet | 710000 | 499.35 MB | 1 |
|
| 656 |
+
| BLKT_corpus-00024.parquet | 468155 | 388.07 MB | 1 |
|
| 657 |
+
|
| 658 |
+
### Loading the dataset / Duomenų rinkinio įkėlimas
|
| 659 |
+
|
| 660 |
+
Example using the `datasets` library:
|
| 661 |
+
|
| 662 |
+
```python
|
| 663 |
+
from datasets import load_dataset
|
| 664 |
+
|
| 665 |
+
ds = load_dataset("parquet", data_files="hf://datasets/VSSA-SDSA/lithuanian-general-corpus/data/*.parquet")
|
| 666 |
+
print(ds)
|
| 667 |
+
```
|
| 668 |
+
|
| 669 |
+
Example using pandas:
|
| 670 |
+
|
| 671 |
+
```python
|
| 672 |
+
import pandas as pd
|
| 673 |
+
|
| 674 |
+
# Example for one file
|
| 675 |
+
df = pd.read_parquet("data/your_file.parquet")
|
| 676 |
+
print(df.head())
|
| 677 |
+
```
|
| 678 |
+
|
| 679 |
+
### Files with metadata read errors / Failai su metaduomenų nuskaitymo klaidomis
|
| 680 |
+
|
| 681 |
+
None.
|
| 682 |
+
<a id=lt></a>
|
| 683 |
+
## LT
|
| 684 |
+
|
| 685 |
+
### Aprašas
|
| 686 |
+
|
| 687 |
+
Bendrasis lietuvių kalbos tekstynas (toliau – BLKT) sukurtas Valstybės skaitmeninių sprendimų agentūrai (VSSA) įgyvendinant projektą „Bendrojo lietuvių kalbos tekstyno ir vektorizuotų modelių sukūrimas“ (projekto kodas: 02-103-P-0001). BLKT sukūrė keturių partnerių konsorciumas: Vytauto Didžiojo universitetas (VDU), UAB „Neurotechnology“, UAB „Tilde informacinės technologijos“ ir MB „Krilas“.
|
| 688 |
+
|
| 689 |
+
Projektas vykdomas vadovaujantis 2021–2030 metų Lietuvos Respublikos valstybės skaitmeninimo plėtros programos pažangos priemone Nr. 05-002-01-07-08 „Kurti technologinius sprendimus ir įrankius, leidžiančius saugiai ir patogiai naudotis paslaugomis“ ir atliepia jos veiklos „Kalbinių išteklių dirbtinio intelekto technologijų sprendimų poreikiams plėtra“ reikalavimus.
|
| 690 |
+
|
| 691 |
+
### Duomenų rinkinio santrauka
|
| 692 |
+
|
| 693 |
+
- Žodžių skaičius: **3 941 476 219**
|
| 694 |
+
- Bendras eilučių (tekstų) skaičius: **8 438 155**
|
| 695 |
+
- Parquet failų skaičius: **25**
|
| 696 |
+
- Bendras failų dydis: **17,79 GB**
|
| 697 |
+
|
| 698 |
+
### Pagrindiniai stulpeliai
|
| 699 |
+
|
| 700 |
+
- `alpha_word_count`: `int64`
|
| 701 |
+
- `author`: `string`
|
| 702 |
+
- `document_subtype`: `string`
|
| 703 |
+
- `document_type`: `string`
|
| 704 |
+
- `id`: `string`
|
| 705 |
+
- `language`: `string`
|
| 706 |
+
- `license`: `string`
|
| 707 |
+
- `period`: `string`
|
| 708 |
+
- `publication_date`: `string`
|
| 709 |
+
- `record_created`: `string`
|
| 710 |
+
- `source_file`: `string`
|
| 711 |
+
- `source_id`: `string`
|
| 712 |
+
- `source_name`: `string`
|
| 713 |
+
- `text`: `string`
|
| 714 |
+
- `text_char_count`: `string`
|
| 715 |
+
- `title`: `null`
|
| 716 |
+
- `url`: `string`
|
| 717 |
+
|
| 718 |
+
### Kilmė ir sudarymas
|
| 719 |
+
|
| 720 |
+
Visi BLKT tekstai skirstomi į 5 teksto tipus (document_types): grožinius (gro), negrožinius (neg), žiniasklaidos (zin), sakytinės kalbos (sak) ir dokumentus (dok). Teksto tipai toliau skirstomi į potipius (subtypes) (žr. lentelę).
|
| 721 |
+
|
| 722 |
+
| Tipas | Apibūdinimas | Potipis | Apibūdinimas |
|
| 723 |
+
| ----- | ------------------------ | ------- | ------------------------------------- |
|
| 724 |
+
| zin | Žiniasklaidos tekstai | port | Interneto naujienų portalai |
|
| 725 |
+
| zin | Žiniasklaidos tekstai | prdk | Laikraščiai, žurnalai |
|
| 726 |
+
| sak | Sakytinės kalbos tekstai | sste | Seimo stenogramos |
|
| 727 |
+
| sak | Sakytinės kalbos tekstai | skit | Kitos sakytinės kalbos transkripcijos |
|
| 728 |
+
| neg | Negrožiniai tekstai | intt | Interneto tekstai |
|
| 729 |
+
| neg | Negrožiniai tekstai | moks | Mokslo straipsniai, knygos |
|
| 730 |
+
| neg | Negrožiniai tekstai | kiti | Kiti tekstai |
|
| 731 |
+
| gro | Grožiniai tekstai | proz | Proza |
|
| 732 |
+
| gro | Grožiniai tekstai | poez | Poezija |
|
| 733 |
+
| dok | Dokumentai | dkes | ES dokumentai |
|
| 734 |
+
| dok | Dokumentai | dklt | LR teisiniai dokumentai |
|
| 735 |
+
| dok | Dokumentai | dkad | Administraciniai dokumentai |
|
| 736 |
+
|
| 737 |
+
Teksto tipų pasiskirstymas:
|
| 738 |
+
|
| 739 |
+
| Teksto tipas | Žodžių sk. | Dalis | Tekstų sk. |
|
| 740 |
+
| --------------------- | -------------: | ----------: | ----------: |
|
| 741 |
+
| Dokumentai | 1509347610 | 38.29% | 1212131 |
|
| 742 |
+
| Grožiniai tekstai | 11767755 | 0.30% | 440 |
|
| 743 |
+
| Negrožiniai tekstai | 279251467 | 7.08% | 832256 |
|
| 744 |
+
| Sakytinės k. tekstai | 80483739 | 2.04% | 4563 |
|
| 745 |
+
| Žiniasklaidos tekstai | 2060625648 | 52.28% | 6388765 |
|
| 746 |
+
| **Iš viso** | **3941476219** | **100.00%** | **8438155** |
|
| 747 |
+
|
| 748 |
+
Tekstai surinkti iš 36 skirtingų šaltinių užtikrinant skaidrumą, iš duomenų tiekėjų gavus visus reikalingus sutikimus ir licencijas.
|
| 749 |
+
|
| 750 |
+
| Nr | Šaltinis | Tekstų skaičius | Žodžių skaičius | Dalis (%) |
|
| 751 |
+
| -: | ----------------------------- | --------------: | --------------: | ----------: |
|
| 752 |
+
| 1 | 15min.lt | 559107 | 230933602 | 5,86% |
|
| 753 |
+
| 2 | autogidas.lt | 1846 | 735416 | 0,02% |
|
| 754 |
+
| 3 | bernardinai.lt | 114040 | 65085920 | 1,65% |
|
| 755 |
+
| 4 | BNS | 352946 | 83877818 | 2,13% |
|
| 756 |
+
| 5 | CRIS-DB | 6483 | 34630986 | 0,88% |
|
| 757 |
+
| 6 | delfi.lt | 1714345 | 606662831 | 15,39% |
|
| 758 |
+
| 7 | elta.lt | 1235494 | 254356849 | 6,45% |
|
| 759 |
+
| 8 | epaveldas.lt | 265 | 11339315 | 0,29% |
|
| 760 |
+
| 9 | etalpykla.vilniustech.lt.lt | 258 | 210746 | 0,01% |
|
| 761 |
+
| 10 | e-tar.lt | 307223 | 567471590 | 14,40% |
|
| 762 |
+
| 11 | eur-lex.europa.eu | 141887 | 319901553 | 8,12% |
|
| 763 |
+
| 12 | Gargždapilis | 99 | 98983 | 0,00% |
|
| 764 |
+
| 13 | gazas.lt | 5454 | 2022545 | 0,05% |
|
| 765 |
+
| 14 | gov.lt | 515173 | 162776172 | 4,13% |
|
| 766 |
+
| 15 | Hubris | 16 | 1118951 | 0,03% |
|
| 767 |
+
| 16 | Humanitarų meka | 64 | 68586 | 0,00% |
|
| 768 |
+
| 17 | Internet archive | 112 | 5592994 | 0,14% |
|
| 769 |
+
| 18 | Įv. autoriai | 182 | 1226753 | 0,03% |
|
| 770 |
+
| 19 | kaunas.kasvyksta.lt | 49662 | 15141979 | 0,38% |
|
| 771 |
+
| 20 | Lituanistika_DB | 286 | 1036134 | 0,03% |
|
| 772 |
+
| 21 | LLTI | 2 | 867891 | 0,02% |
|
| 773 |
+
| 22 | lrytas.lt | 1140602 | 380082600 | 9,64% |
|
| 774 |
+
| 23 | lrkt.lt | 1029 | 4162009 | 0,11% |
|
| 775 |
+
| 24 | lrs.lt | 134968 | 118782124 | 3,01% |
|
| 776 |
+
| 25 | lrt.lt | 761744 | 244438996 | 6,20% |
|
| 777 |
+
| 26 | MIDAS | 42 | 39747 | 0,00% |
|
| 778 |
+
| 27 | pirkimai.eviesiejipirkimai.lt | 46288 | 117844194 | 2,99% |
|
| 779 |
+
| 28 | raudonoji_knyga | 396 | 103296 | 0,00% |
|
| 780 |
+
| 29 | sena.lt | 88 | 31883 | 0,00% |
|
| 781 |
+
| 30 | Švenčionių kraštas | 11046 | 5253903 | 0,13% |
|
| 782 |
+
| 31 | tautinispaveldas.lt | 392 | 59793 | 0,00% |
|
| 783 |
+
| 32 | teismai.lt | 715704 | 499968264 | 12,68% |
|
| 784 |
+
| 33 | vdu.lt | 5937 | 709234 | 0,02% |
|
| 785 |
+
| 34 | ve.lt | 427947 | 160571897 | 4,07% |
|
| 786 |
+
| 35 | Vikipedija | 170718 | 34741743 | 0,88% |
|
| 787 |
+
| 36 | zmones.15min.lt | 16310 | 9528922 | 0,24% |
|
| 788 |
+
| | **Iš viso** | **8438155** | **3941476219** | **100,00%** |
|
| 789 |
+
|
| 790 |
+
Surinkti tekstai publikuoti
|
| 791 |
+
|
| 792 |
+
| Laikotarpis | Žyma | Žodžių sk. | Dalis | Tekstų sk. |
|
| 793 |
+
| ----------- | ----------: | -------------: | ----------: | ----------: |
|
| 794 |
+
| 1922–1940 | 1 | 11226438 | 0,28% | 290 |
|
| 795 |
+
| 1941–1990 | 2 | 10440807 | 0,26% | 4288 |
|
| 796 |
+
| 1990–2004 | 3 | 229775041 | 5,83% | 684357 |
|
| 797 |
+
| 2008–2026 | 4 | 3690033933 | 93,62% | 7749220 |
|
| 798 |
+
| | **Iš viso** | **3941476219** | **100,00%** | **8438155** |
|
| 799 |
+
|
| 800 |
+
|
| 801 |
+
### Licencija
|
| 802 |
+
|
| 803 |
+
Šis duomenų rinkinys yra platinamas pagal **NewGenLTU OpenRAIL-D** licenciją.
|
| 804 |
+
|
| 805 |
+
Ši „Open RAIL-D“ licencija buvo specialiai sukurta Vytauto Didžiojo universiteto tyrėjų komandos, siekiant skelbti projekto rezultatus – duomenų rinkinius, – kurie buvo parengti vykdant mokslinių tyrimų ir plėtros projektus, finansuojamus pagal programą „Naujos kartos Lietuvoje“. Programa „Naujos kartos Lietuva“ finansuojama iš Ekonomikos gaivinimo ir atsparumo didinimo priemonės (RRF). Buvo nuspręsta sukurti šią licenciją siekiant pašalinti tam tikrus duomenų rinkinių naudojimo apribojimus ir užtikrinti atitiktį ES AI aktui, duomenų apsaugos reikalavimams bei autorių teisių įstatymams.
|
| 806 |
+
|
| 807 |
+
Šia licencija siekiama atviro ir atsakingo modelio ir (arba) duomenų rinkinių tolesnio naudojimo, tuo pačiu užtikrinant, kad duomenų rinkinyje esantys kūriniai, kuriems taikomos autorių teisės, nebūtų naudojami kūriniams atkurti, o būtų naudojami AI mokymui ir kitai technologijų plėtros veiklai. Panašiai, kaip ir modelių atveju, licencija siekia užtikrinti atitiktį ES duomenų apsaugos reglamentui, taigi draudžia asmens duomenų atkūrimą ar išgavimą iš duomenų rinkinio.
|
| 808 |
+
|
| 809 |
+
Ši licencija reglamentuoja duomenų rinkinio ir (arba) modelio (ir jų išvestinių kūrinių) naudojimą.
|
| 810 |
+
Prašome peržiūrėti pridedamą failą „LICENSE.txt“ šiame saugykloje.
|
| 811 |
+
|
| 812 |
+
### Rekomenduojamas naudojimas
|
| 813 |
+
|
| 814 |
+
- Generatyvinė užduotis: automatinį tekstų kūrimas, apibendrinimų generavimas, kūrybinių idėjų ir scenarijų generavimas.
|
| 815 |
+
- Redagavimo užduotis ir korektūra: gramatikos, stiliaus teksto aiškumo ir struktūros tobulinimas.
|
| 816 |
+
- Verslo sprendimai:
|
| 817 |
+
- Kalbos redagavimo įrankiai
|
| 818 |
+
- Išmanioji paieška ir teksto santraukų generavimas.
|
| 819 |
+
- Virtualių asistentų, rinkodaros tekstų kūrimas.
|
| 820 |
+
- Duomenų analizė ir klientų poreikių įžvalgos.
|
| 821 |
+
|
| 822 |
+
### Apribojimai
|
| 823 |
+
|
| 824 |
+
Pagal **NewGenLTU OpenRAIL-D** licenciją vartotojai negali naudoti šio duomenų rinkinio šiais tikslais: diskriminacijai; ginklų ar karinių programų kūrimui; automatizuotam sprendimų priėmimui; dezinformacijos skleidimui; privatumo pažeidimams; medicininių ar sveikatos patarimų teikimui; kenkėjiškų programų kūrimui; priekabiavimui; nesąžiningiems tyrimams; arba asmens duomenų rinkimui.
|
| 825 |
+
Daugiau informacijos rasite šioje saugykloje esančiame pridedamame faile `LICENSE.txt`.
|
| 826 |
+
|
| 827 |
+
### Ribotumai
|
| 828 |
+
|
| 829 |
+
Kūrėjai dėjo visas pastangas, kad tekstyną išvalytų nuo šiukšlių, klaidų ir dubliažų. Tačiau:
|
| 830 |
+
- Rašybos klaidų dalis: 0,46%
|
| 831 |
+
- Kitų kalbų žodžių dalis: 0,07%
|
| 832 |
+
- 1–2 laikotarpių tekstuose (0,54 %) gali būti senų, nebevartojamų žodžių, frazių ar posakių
|
| 833 |
+
- Tekstye vyrauja tekstai iš naujienų portalų (52 %) ir dokumentų (38 %).
|
| 834 |
+
|
| 835 |
+
### Citavimas
|
| 836 |
+
|
| 837 |
+
"Bendrasis lietuvių kalbos tekstynas".
|
| 838 |
+
|
| 839 |
+
|