Update README.md
Browse files
README.md
CHANGED
|
@@ -5,10 +5,51 @@ task_categories:
|
|
| 5 |
language:
|
| 6 |
- my
|
| 7 |
tags:
|
| 8 |
-
- sentiment
|
| 9 |
- myanmar
|
| 10 |
- burmese
|
| 11 |
pretty_name: Myanmar Sentiment Intensity Dataset v1
|
| 12 |
size_categories:
|
| 13 |
- 1K<n<10K
|
| 14 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5 |
language:
|
| 6 |
- my
|
| 7 |
tags:
|
| 8 |
+
- sentiment
|
| 9 |
- myanmar
|
| 10 |
- burmese
|
| 11 |
pretty_name: Myanmar Sentiment Intensity Dataset v1
|
| 12 |
size_categories:
|
| 13 |
- 1K<n<10K
|
| 14 |
+
---
|
| 15 |
+
|
| 16 |
+
# Myanmar Sentiment Intensity Dataset v1
|
| 17 |
+
|
| 18 |
+
This dataset contains Myanmar (Burmese) text samples annotated for sentiment intensity across 5 levels. It is specifically curated for training and evaluating text classification models.
|
| 19 |
+
|
| 20 |
+
## Dataset Structure
|
| 21 |
+
The dataset consists of two primary columns:
|
| 22 |
+
- **text**: Raw Myanmar text strings.
|
| 23 |
+
- **label**: Sentiment intensity score (Integer 0 to 4).
|
| 24 |
+
|
| 25 |
+
### Label Definitions
|
| 26 |
+
| Label | Emoji | Sentiment | Description |
|
| 27 |
+
|---|---|---|---|
|
| 28 |
+
| 0 | 🔴 | Very Negative | Strong dissatisfaction, anger, or hate |
|
| 29 |
+
| 1 | 🟠 | Negative | Dissatisfaction, boredom, or sadness |
|
| 30 |
+
| 2 | 🟡 | Neutral | Normal, factual, or indifferent |
|
| 31 |
+
| 3 | 🟢 | Positive | Satisfied, good, or peaceful |
|
| 32 |
+
| 4 | 🔵 | Very Positive | Very happy, excellent, or joyful |
|
| 33 |
+
|
| 34 |
+
## Data Preprocessing
|
| 35 |
+
To achieve optimal results with this dataset, it is highly recommended to perform **Syllable Breaking** before tokenization. This ensures the model processes the text using Myanmar syllable units.
|
| 36 |
+
|
| 37 |
+
### Syllable Breaking Logic:
|
| 38 |
+
The regex implementation is based on the [ye-kyaw-thu/sylbreak](https://github.com/ye-kyaw-thu/sylbreak) algorithm.
|
| 39 |
+
|
| 40 |
+
```python
|
| 41 |
+
import re
|
| 42 |
+
|
| 43 |
+
def myanmar_sylbreak(line):
|
| 44 |
+
pat = re.compile(r"((?<!္)[က-အ](?![်္])|[a-zA-Z0-9ဣဤဥဦဧဩဪဿ၌၍၏၀-၉၊။!-/:-@[-`{-~\s])")
|
| 45 |
+
line = re.sub(r'\s+', ' ', str(line).strip())
|
| 46 |
+
return pat.sub(r" \1", line).strip()
|
| 47 |
+
|
| 48 |
+
```
|
| 49 |
+
|
| 50 |
+
## Acknowledgments & Credits
|
| 51 |
+
|
| 52 |
+
This dataset is developed by building upon existing Myanmar NLP resources and research:
|
| 53 |
+
|
| 54 |
+
* **Base Dataset Source:** This work utilizes data and concepts from the [Myanmar Text Segmentation Dataset](https://huggingface.co/datasets/chuuhtetnaing/myanmar-social-media-sentiment-analysis-dataset) by Chuu Htet Naing.
|
| 55 |
+
* **Syllable Segmentation:** The syllable breaking logic is based on the [sylbreak](https://github.com/ye-kyaw-thu/sylbreak) repository by Ye Kyaw Thu.
|