|
|
--- |
|
|
license: mit |
|
|
language: |
|
|
- en |
|
|
base_model: |
|
|
- google-bert/bert-base-uncased |
|
|
tags: |
|
|
- cybersecurity |
|
|
--- |
|
|
# Cybersecurity Aspect Term Extraction and Polarity Classification |
|
|
|
|
|
## Model Description |
|
|
This model is trained to extract aspect terms and classify their sentiment polarity in cybersecurity texts. |
|
|
|
|
|
## Model Architecture |
|
|
- **Base Model**: BERT-base-uncased |
|
|
- **Architecture**: FAST_LCF_ATEPC Fast Local Content Focus-Aspect Term Extraction Polarity Classification |
|
|
|
|
|
## Performance Metrics |
|
|
- **APC F1**: 41.24% |
|
|
- **ATE F1**: 90.57% |
|
|
- **APC Accuracy**: 65.23% |
|
|
|
|
|
## Usage |
|
|
```python |
|
|
from pyabsa import AspectTermExtraction as ATEPC |
|
|
|
|
|
# Load the model |
|
|
aspect_extractor = ATEPC.AspectExtractor(checkpoint="adoamesh/PyABSA_Cybersecurity_ATE_Polarity_Classification") |
|
|
|
|
|
# Predict aspects and sentiments |
|
|
result = aspect_extractor.predict("A ransomware attack targeted the hospital's patient records system.") |
|
|
print(result) |
|
|
``` |
|
|
|
|
|
## Training Data |
|
|
The model was trained on a custom cybersecurity dataset with IOB format annotations. |
|
|
|
|
|
## Limitations |
|
|
This model is trained on cybersecurity texts and may not perform well on other domains. |
|
|
|
|
|
## Biases |
|
|
The model may reflect biases present in the training data. |
|
|
|
|
|
## License |
|
|
MIT |
|
|
|
|
|
## Author |
|
|
Daniel Amemba Odhiambo |
