Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,45 @@
|
|
1 |
---
|
2 |
license: unknown
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: unknown
|
3 |
---
|
4 |
+
|
5 |
+
## Openrice Sentiment Classification dataset
|
6 |
+
|
7 |
+
From [github.com/Christainx/Dataset_Cantonese_Openrice](https://github.com/Christainx/Dataset_Cantonese_Openrice).
|
8 |
+
|
9 |
+
The dataset includes 60k instances from Cantonese reviews in Openrice.
|
10 |
+
|
11 |
+
The rating ranks from 1-star (very negative) to 5-star (very positive). The instances are shuffled in order to disperse reviews of same restaurant.
|
12 |
+
|
13 |
+
### Code for the splits creation
|
14 |
+
|
15 |
+
```
|
16 |
+
import datasets
|
17 |
+
|
18 |
+
def load_openrice():
|
19 |
+
# https://github.com/Christainx/Dataset_Cantonese_Openrice/blob/master/Openrice_Cantonese.7z
|
20 |
+
with open('Openrice_Cantonese.txt') as file:
|
21 |
+
for i, line in enumerate(file):
|
22 |
+
label = int(line[0])
|
23 |
+
text = line[1:].strip()
|
24 |
+
yield {'text': text, 'label': label}
|
25 |
+
|
26 |
+
|
27 |
+
ds = datasets.Dataset.from_generator(load_openrice)
|
28 |
+
print(ds)
|
29 |
+
dsd = ds.train_test_split(0.2, seed=42)
|
30 |
+
dsd['test'].to_json('data/test.jsonl', orient='records', force_ascii=False)
|
31 |
+
dsd['train'].to_json('data/train.jsonl', orient='records', force_ascii=False)
|
32 |
+
print(dsd)
|
33 |
+
```
|
34 |
+
|
35 |
+
## Citation
|
36 |
+
```
|
37 |
+
@inproceedings{xiang2019sentiment,
|
38 |
+
title={Sentiment Augmented Attention Network for Cantonese Restaurant Review Analysis},
|
39 |
+
author={Xiang, Rong and Jiao, Ying and Lu, Qin},
|
40 |
+
booktitle={Proceedings of the 8th KDD Workshop on Issues of Sentiment Discovery and Opinion Mining (WISDOM)},
|
41 |
+
pages={1--9},
|
42 |
+
year={2019},
|
43 |
+
organization={KDD WISDOM}
|
44 |
+
}
|
45 |
+
```
|