Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,129 @@
|
|
1 |
---
|
2 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
+
annotations_creators:
|
3 |
+
- crowdsourced
|
4 |
+
language_creators:
|
5 |
+
- crowdsourced
|
6 |
+
languages:
|
7 |
+
- uk-UA
|
8 |
+
licenses:
|
9 |
+
- cc-by-3.0
|
10 |
+
multilinguality:
|
11 |
+
- monolingual
|
12 |
+
- translation
|
13 |
+
pretty_name: 'Ukrainian Wikipedia edits '
|
14 |
+
size_categories:
|
15 |
+
- 1M<n<10M
|
16 |
+
source_datasets:
|
17 |
+
- original
|
18 |
+
task_categories:
|
19 |
+
- other
|
20 |
+
task_ids: []
|
21 |
---
|
22 |
+
|
23 |
+
# Ukrainian Wikipedia Edits
|
24 |
+
|
25 |
+
### Dataset summary
|
26 |
+
|
27 |
+
A corpus of sentence edits extracted from Ukrainian Wikipedia history revisions.
|
28 |
+
|
29 |
+
Edits were filtered by edit distance and sentence length. This makes them usable for grammatical error correction (GEC) or spellchecker models pre-training.
|
30 |
+
|
31 |
+
|
32 |
+
### Supported Tasks and Leaderboards
|
33 |
+
|
34 |
+
* Ukrainian grammatical error correction (GEC) - see [UA-GEC](https://github.com/grammarly/ua-gec)
|
35 |
+
* Ukrainian spelling correction
|
36 |
+
|
37 |
+
### Languages
|
38 |
+
|
39 |
+
Ukrainian
|
40 |
+
|
41 |
+
## Dataset Structure
|
42 |
+
|
43 |
+
### Data Fields
|
44 |
+
|
45 |
+
* `src` - sentence before edit
|
46 |
+
* `tgt` - sentence after edit
|
47 |
+
|
48 |
+
### Data Splits
|
49 |
+
|
50 |
+
* `full/train` contains all the data
|
51 |
+
* `tiny/train` contains a 5000 examples sample.
|
52 |
+
|
53 |
+
## Dataset Creation
|
54 |
+
|
55 |
+
Latest full Ukrainian Wiki dump were used as of 2022-04-30.
|
56 |
+
|
57 |
+
It was processed with the [wikiedits](https://github.com/snukky/wikiedits) and custom scripts.
|
58 |
+
|
59 |
+
### Source Data
|
60 |
+
|
61 |
+
#### Initial Data Collection and Normalization
|
62 |
+
|
63 |
+
Wikipedia
|
64 |
+
|
65 |
+
#### Who are the source language producers?
|
66 |
+
|
67 |
+
Wikipedia writers
|
68 |
+
|
69 |
+
### Annotations
|
70 |
+
|
71 |
+
#### Annotation process
|
72 |
+
|
73 |
+
Annotations inferred by comparing two subsequent page revisions.
|
74 |
+
|
75 |
+
#### Who are the annotators?
|
76 |
+
|
77 |
+
People who edit Wikipedia pages.
|
78 |
+
|
79 |
+
### Personal and Sensitive Information
|
80 |
+
|
81 |
+
No
|
82 |
+
|
83 |
+
## Considerations for Using the Data
|
84 |
+
|
85 |
+
### Social Impact of Dataset
|
86 |
+
|
87 |
+
[More Information Needed]
|
88 |
+
|
89 |
+
### Discussion of Biases
|
90 |
+
|
91 |
+
[More Information Needed]
|
92 |
+
|
93 |
+
### Other Known Limitations
|
94 |
+
|
95 |
+
The data is noisy. In addition to GEC and spelling edits, it contains a good chunk of factual changes and vandalism.
|
96 |
+
|
97 |
+
More task-specific filters could help.
|
98 |
+
|
99 |
+
## Additional Information
|
100 |
+
|
101 |
+
### Dataset Curators
|
102 |
+
|
103 |
+
[Oleksiy Syvokon](https://github.com/asivokon)
|
104 |
+
|
105 |
+
### Licensing Information
|
106 |
+
|
107 |
+
CC-BY-3.0
|
108 |
+
|
109 |
+
### Citation Information
|
110 |
+
|
111 |
+
```
|
112 |
+
@inproceedings{wiked2014,
|
113 |
+
author = {Roman Grundkiewicz and Marcin Junczys-Dowmunt},
|
114 |
+
title = {The WikEd Error Corpus: A Corpus of Corrective Wikipedia Edits and its Application to Grammatical Error Correction},
|
115 |
+
booktitle = {Advances in Natural Language Processing -- Lecture Notes in Computer Science},
|
116 |
+
editor = {Adam Przepiórkowski and Maciej Ogrodniczuk},
|
117 |
+
publisher = {Springer},
|
118 |
+
year = {2014},
|
119 |
+
volume = {8686},
|
120 |
+
pages = {478--490},
|
121 |
+
url = {http://emjotde.github.io/publications/pdf/mjd.poltal2014.draft.pdf}
|
122 |
+
}
|
123 |
+
```
|
124 |
+
|
125 |
+
### Contributions
|
126 |
+
|
127 |
+
[@snukky](https://github.com/snukky) created tools for dataset processing.
|
128 |
+
|
129 |
+
[@asivokon](https://github.com/asivokon) generated this dataset.
|