Update classifier_manager/deberta_model.py
Browse files
classifier_manager/deberta_model.py
CHANGED
|
@@ -26,4 +26,33 @@ class PiiDebertaAnalyzer:
|
|
| 26 |
print(f"✅ DeBERTa model '{model_name}' loaded successfully.")
|
| 27 |
|
| 28 |
except Exception as e:
|
| 29 |
-
print(f"Failed to load DeBERTa model: {e}")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 26 |
print(f"✅ DeBERTa model '{model_name}' loaded successfully.")
|
| 27 |
|
| 28 |
except Exception as e:
|
| 29 |
+
print(f"Failed to load DeBERTa model: {e}")
|
| 30 |
+
|
| 31 |
+
def scan(self, text: str):
|
| 32 |
+
if not self.model_loaded or not text:
|
| 33 |
+
return []
|
| 34 |
+
|
| 35 |
+
try:
|
| 36 |
+
results = self.pipe(text)
|
| 37 |
+
detections = []
|
| 38 |
+
|
| 39 |
+
for entity in results:
|
| 40 |
+
# entity looks like: {'entity_group': 'NAME_STUDENT', 'score': 0.99, 'word': 'John Doe', 'start': 0, 'end': 8}
|
| 41 |
+
original_label = entity.get('entity_group', 'UNKNOWN')
|
| 42 |
+
mapped_label = self.label_mapping.get(original_label, "DEFAULT")
|
| 43 |
+
|
| 44 |
+
# Only include known PII types
|
| 45 |
+
if mapped_label != "DEFAULT":
|
| 46 |
+
detections.append({
|
| 47 |
+
"text": entity['word'].strip(),
|
| 48 |
+
"label": mapped_label,
|
| 49 |
+
"start": entity['start'],
|
| 50 |
+
"end": entity['end'],
|
| 51 |
+
"source": "DeBERTa",
|
| 52 |
+
"score": float(entity['score'])
|
| 53 |
+
})
|
| 54 |
+
return detections
|
| 55 |
+
|
| 56 |
+
except Exception as e:
|
| 57 |
+
print(f"DeBERTa scan error: {e}")
|
| 58 |
+
return []
|