Datasets:
Tasks:
Text Classification
Formats:
parquet
Sub-tasks:
sentiment-classification
Languages:
Korean
Size:
100K - 1M
License:
Update README.md
Browse files
README.md
CHANGED
@@ -23,14 +23,6 @@ Korean sentiment classification dataset
|
|
23 |
- Size: 460K(+180K)
|
24 |
- Language: Korean-centric
|
25 |
|
26 |
-
### Usage
|
27 |
-
```python
|
28 |
-
from datasets import load_dataset
|
29 |
-
|
30 |
-
kr3 = load_dataset("Wittgensteinian/KR3", name='kr3', split='train')
|
31 |
-
kr3 = kr3.remove_columns(['__index_level_0__']) # Original file didn't include this column. Suspect it's a hugging face issue.
|
32 |
-
```
|
33 |
-
|
34 |
### ⚠️ Caution with `Rating` Column
|
35 |
0 stands for negative review, 1 stands for positive review, and 2 stands for ambiguous review.
|
36 |
**Note that rating 2 is not intended to be used directly for supervised learning(classification).** This data is included for additional pre-training purpose or other usage.
|
@@ -40,6 +32,19 @@ In other words, this dataset is basically a **binary** sentiment classification
|
|
40 |
See all the codes for crawling/preprocessing the dataset and experiments with KR3 in [Gitlab Repo](https://gitlab.com/Wittgensteinian/kr3).
|
41 |
See Kaggle dataset in [Kaggle Dataset](https://www.kaggle.com/ninetyninenewton/kr3-korean-restaurant-reviews-with-ratings).
|
42 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
43 |
### License
|
44 |
**CC BY-NC-SA 4.0**
|
45 |
|
|
|
23 |
- Size: 460K(+180K)
|
24 |
- Language: Korean-centric
|
25 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
26 |
### ⚠️ Caution with `Rating` Column
|
27 |
0 stands for negative review, 1 stands for positive review, and 2 stands for ambiguous review.
|
28 |
**Note that rating 2 is not intended to be used directly for supervised learning(classification).** This data is included for additional pre-training purpose or other usage.
|
|
|
32 |
See all the codes for crawling/preprocessing the dataset and experiments with KR3 in [Gitlab Repo](https://gitlab.com/Wittgensteinian/kr3).
|
33 |
See Kaggle dataset in [Kaggle Dataset](https://www.kaggle.com/ninetyninenewton/kr3-korean-restaurant-reviews-with-ratings).
|
34 |
|
35 |
+
### Usage
|
36 |
+
```python
|
37 |
+
from datasets import load_dataset
|
38 |
+
|
39 |
+
kr3 = load_dataset("Wittgensteinian/KR3", name='kr3', split='train')
|
40 |
+
kr3 = kr3.remove_columns(['__index_level_0__']) # Original file didn't include this column. Suspect it's a hugging face issue.
|
41 |
+
```
|
42 |
+
```python
|
43 |
+
# drop reviews with ambiguous label
|
44 |
+
kr3_binary = kr3.filter(lambda example: example['Rating'] != 2)
|
45 |
+
```
|
46 |
+
|
47 |
+
|
48 |
### License
|
49 |
**CC BY-NC-SA 4.0**
|
50 |
|