Update README.md
Browse files
README.md
CHANGED
@@ -28,4 +28,20 @@ configs:
|
|
28 |
path: data/dev-*
|
29 |
- split: test
|
30 |
path: data/test-*
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
31 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
28 |
path: data/dev-*
|
29 |
- split: test
|
30 |
path: data/test-*
|
31 |
+
license: cc0-1.0
|
32 |
+
task_categories:
|
33 |
+
- token-classification
|
34 |
+
language:
|
35 |
+
- en
|
36 |
+
tags:
|
37 |
+
- CD-SCO
|
38 |
---
|
39 |
+
|
40 |
+
This dataset is converted from [SEM Shared Task 2012: Resolving the Scope and Focus of Negation](https://www.clips.ua.ac.be/sem2012-st-neg/data.html), [CD-SCO] dataset.
|
41 |
+
|
42 |
+
The creation of this data follows the approach of the Augment method described in [NegBERT (Khandelwal, et al. 2020)](http://www.lrec-conf.org/proceedings/lrec2020/pdf/2020.lrec-1.704.pdf).
|
43 |
+
That is, adding a special token ([NEG]) immediately before the predicate:
|
44 |
+
> This is [NEG] not a sentence.
|
45 |
+
|
46 |
+
Note that **the special token and the predicate is considered a whole**. That is, the actual sentence is like
|
47 |
+
> 'This' 'is' **'[NEG] not'** 'a' 'sentence' '.'
|