Update README.md
Browse files
README.md
CHANGED
|
@@ -20,27 +20,9 @@ widget:
|
|
| 20 |
|
| 21 |
# # Arabic text classification using deep learning (ArabicT5)
|
| 22 |
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
- Paper
|
| 26 |
-
[https://www.researchgate.net/publication/333605992_SANAD_Single-Label_Arabic_News_Articles_Dataset_for_Automatic_Text_Categorization]
|
| 27 |
-
|
| 28 |
-
-Dataset
|
| 29 |
-
[https://data.mendeley.com/datasets/57zpx667y9/2]
|
| 30 |
-
|
| 31 |
-
# # Their experiment'
|
| 32 |
-
|
| 33 |
-
[https://www.sciencedirect.com/science/article/abs/pii/S0306457319303413]
|
| 34 |
-
|
| 35 |
-
"Our experimental results showed that all models did very well on SANAD corpus with a minimum accuracy of 93.43%, achieved by CGRU, and top performance of 95.81%, achieved by HANGRU."
|
| 36 |
-
| Model | Accuracy |
|
| 37 |
-
| :---------------------: | :---------------------: |
|
| 38 |
-
| CGRU | 93.43% |
|
| 39 |
-
| HANGRU | 95.81% |
|
| 40 |
-
|
| 41 |
-
# # Our experiment
|
| 42 |
|
| 43 |
-
|
| 44 |
category_mapping = {
|
| 45 |
|
| 46 |
'Politics':1,
|
|
@@ -52,7 +34,7 @@ widget:
|
|
| 52 |
'Religion':7
|
| 53 |
}
|
| 54 |
|
| 55 |
-
|
| 56 |
|
| 57 |
| | |
|
| 58 |
| :-------------------: | :-----------:|
|
|
@@ -65,7 +47,7 @@ widget:
|
|
| 65 |
| Epoch | `2` |
|
| 66 |
| | |
|
| 67 |
|
| 68 |
-
|
| 69 |
|
| 70 |
| | |
|
| 71 |
| :---------------------: | :-----------: |
|
|
@@ -73,6 +55,27 @@ widget:
|
|
| 73 |
| Accuracy | `96.49%` |
|
| 74 |
| BLeU | `96.49%` |
|
| 75 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 76 |
# # Example usage
|
| 77 |
```python
|
| 78 |
from transformers import T5ForConditionalGeneration, T5Tokenizer
|
|
|
|
| 20 |
|
| 21 |
# # Arabic text classification using deep learning (ArabicT5)
|
| 22 |
|
| 23 |
+
# # Our experiment
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 24 |
|
| 25 |
+
- The category mapping
|
| 26 |
category_mapping = {
|
| 27 |
|
| 28 |
'Politics':1,
|
|
|
|
| 34 |
'Religion':7
|
| 35 |
}
|
| 36 |
|
| 37 |
+
- Training parameters
|
| 38 |
|
| 39 |
| | |
|
| 40 |
| :-------------------: | :-----------:|
|
|
|
|
| 47 |
| Epoch | `2` |
|
| 48 |
| | |
|
| 49 |
|
| 50 |
+
- Results
|
| 51 |
|
| 52 |
| | |
|
| 53 |
| :---------------------: | :-----------: |
|
|
|
|
| 55 |
| Accuracy | `96.49%` |
|
| 56 |
| BLeU | `96.49%` |
|
| 57 |
|
| 58 |
+
# # SANAD: Single-label Arabic News Articles Dataset for automatic text categorization
|
| 59 |
+
|
| 60 |
+
- Paper
|
| 61 |
+
[https://www.researchgate.net/publication/333605992_SANAD_Single-Label_Arabic_News_Articles_Dataset_for_Automatic_Text_Categorization]
|
| 62 |
+
|
| 63 |
+
- Dataset
|
| 64 |
+
[https://data.mendeley.com/datasets/57zpx667y9/2]
|
| 65 |
+
|
| 66 |
+
# # Arabic text classification using deep learning models
|
| 67 |
+
|
| 68 |
+
- Paper
|
| 69 |
+
[https://www.sciencedirect.com/science/article/abs/pii/S0306457319303413]
|
| 70 |
+
|
| 71 |
+
- Their experiment'
|
| 72 |
+
"Our experimental results showed that all models did very well on SANAD corpus with a minimum accuracy of 93.43%, achieved by CGRU, and top performance of 95.81%, achieved by HANGRU."
|
| 73 |
+
| Model | Accuracy |
|
| 74 |
+
| :---------------------: | :---------------------: |
|
| 75 |
+
| CGRU | 93.43% |
|
| 76 |
+
| HANGRU | 95.81% |
|
| 77 |
+
|
| 78 |
+
|
| 79 |
# # Example usage
|
| 80 |
```python
|
| 81 |
from transformers import T5ForConditionalGeneration, T5Tokenizer
|