Update app.py
Browse files
app.py
CHANGED
|
@@ -15,19 +15,46 @@ class AnonymizationEvaluator:
|
|
| 15 |
self.results_df = None
|
| 16 |
|
| 17 |
def extract_entities_from_text(self, text: str) -> Dict[str, Set[str]]:
|
| 18 |
-
"""استخراج موجودیتها از متن"""
|
| 19 |
if pd.isna(text) or not isinstance(text, str):
|
| 20 |
return {'companies': set(), 'persons': set(), 'amounts': set(), 'percents': set(), 'groups': set()}
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
|
| 26 |
-
|
| 27 |
-
'
|
|
|
|
|
|
|
|
|
|
|
|
|
| 28 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 29 |
return entities
|
| 30 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 31 |
def calculate_precision_recall_f1(self, reference_entities: Dict[str, Set[str]],
|
| 32 |
predicted_entities: Dict[str, Set[str]]) -> Tuple[float, float, float]:
|
| 33 |
"""محاسبه Precision, Recall و F1-Score"""
|
|
@@ -101,7 +128,7 @@ class AnonymizationEvaluator:
|
|
| 101 |
return 0.0, 0.0, 0.0
|
| 102 |
|
| 103 |
def evaluate_dataset(self, file_path: str) -> Tuple[bool, str, pd.DataFrame]:
|
| 104 |
-
"""ارزیابی کل دیتاست"""
|
| 105 |
try:
|
| 106 |
# بارگذاری فایل
|
| 107 |
df = pd.read_csv(file_path)
|
|
@@ -113,11 +140,27 @@ class AnonymizationEvaluator:
|
|
| 113 |
if missing_columns:
|
| 114 |
return False, f"ستونهای مفقود: {', '.join(missing_columns)}", pd.DataFrame()
|
| 115 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 116 |
# محاسبه متریکها برای هر سطر
|
| 117 |
precisions = []
|
| 118 |
recalls = []
|
| 119 |
f1_scores = []
|
| 120 |
|
|
|
|
|
|
|
| 121 |
for index, row in df.iterrows():
|
| 122 |
precision, recall, f1 = self.evaluate_single_row(
|
| 123 |
row['Reference_text'],
|
|
@@ -127,6 +170,12 @@ class AnonymizationEvaluator:
|
|
| 127 |
precisions.append(round(precision, 4))
|
| 128 |
recalls.append(round(recall, 4))
|
| 129 |
f1_scores.append(round(f1, 4))
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 130 |
|
| 131 |
# اضافه کردن ستونهای جدید
|
| 132 |
df['Precision'] = precisions
|
|
@@ -136,7 +185,12 @@ class AnonymizationEvaluator:
|
|
| 136 |
# ذخیره نتایج
|
| 137 |
self.results_df = df
|
| 138 |
|
| 139 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 140 |
|
| 141 |
except Exception as e:
|
| 142 |
return False, f"خطا در پردازش فایل: {str(e)}", pd.DataFrame()
|
|
|
|
| 15 |
self.results_df = None
|
| 16 |
|
| 17 |
def extract_entities_from_text(self, text: str) -> Dict[str, Set[str]]:
|
| 18 |
+
"""استخراج موجودیتها از متن با debugging"""
|
| 19 |
if pd.isna(text) or not isinstance(text, str):
|
| 20 |
return {'companies': set(), 'persons': set(), 'amounts': set(), 'percents': set(), 'groups': set()}
|
| 21 |
+
|
| 22 |
+
# تمیز کردن متن
|
| 23 |
+
text = str(text).strip()
|
| 24 |
+
|
| 25 |
+
# الگوهای مختلف برای موجودیتها
|
| 26 |
+
patterns = {
|
| 27 |
+
'companies': [r'company-(\d+)', r'Company-(\d+)', r'COMPANY-(\d+)'],
|
| 28 |
+
'persons': [r'person-(\d+)', r'Person-(\d+)', r'PERSON-(\d+)'],
|
| 29 |
+
'amounts': [r'amount-(\d+)', r'Amount-(\d+)', r'AMOUNT-(\d+)'],
|
| 30 |
+
'percents': [r'percent-(\d+)', r'Percent-(\d+)', r'PERCENT-(\d+)'],
|
| 31 |
+
'groups': [r'group-(\d+)', r'Group-(\d+)', r'GROUP-(\d+)']
|
| 32 |
}
|
| 33 |
+
|
| 34 |
+
entities = {}
|
| 35 |
+
for entity_type, pattern_list in patterns.items():
|
| 36 |
+
found = set()
|
| 37 |
+
for pattern in pattern_list:
|
| 38 |
+
matches = re.findall(pattern, text)
|
| 39 |
+
found.update(matches)
|
| 40 |
+
entities[entity_type] = found
|
| 41 |
+
|
| 42 |
return entities
|
| 43 |
|
| 44 |
+
def debug_text_analysis(self, reference_text: str, predicted_text: str, row_num: int = 0) -> str:
|
| 45 |
+
"""تابع debugging برای تحلیل متنها"""
|
| 46 |
+
debug_info = f"\n--- Debug Row {row_num + 1} ---\n"
|
| 47 |
+
debug_info += f"Reference: '{reference_text[:100]}...'\n"
|
| 48 |
+
debug_info += f"Predicted: '{predicted_text[:100]}...'\n"
|
| 49 |
+
|
| 50 |
+
ref_entities = self.extract_entities_from_text(reference_text)
|
| 51 |
+
pred_entities = self.extract_entities_from_text(predicted_text)
|
| 52 |
+
|
| 53 |
+
debug_info += f"Reference entities: {dict(ref_entities)}\n"
|
| 54 |
+
debug_info += f"Predicted entities: {dict(pred_entities)}\n"
|
| 55 |
+
|
| 56 |
+
return debug_info
|
| 57 |
+
|
| 58 |
def calculate_precision_recall_f1(self, reference_entities: Dict[str, Set[str]],
|
| 59 |
predicted_entities: Dict[str, Set[str]]) -> Tuple[float, float, float]:
|
| 60 |
"""محاسبه Precision, Recall و F1-Score"""
|
|
|
|
| 128 |
return 0.0, 0.0, 0.0
|
| 129 |
|
| 130 |
def evaluate_dataset(self, file_path: str) -> Tuple[bool, str, pd.DataFrame]:
|
| 131 |
+
"""ارزیابی کل دیتاست با debugging"""
|
| 132 |
try:
|
| 133 |
# بارگذاری فایل
|
| 134 |
df = pd.read_csv(file_path)
|
|
|
|
| 140 |
if missing_columns:
|
| 141 |
return False, f"ستونهای مفقود: {', '.join(missing_columns)}", pd.DataFrame()
|
| 142 |
|
| 143 |
+
# تشخیص مشکل - بررسی نمونهای از دادهها
|
| 144 |
+
debug_info = "\n=== Debug Information ===\n"
|
| 145 |
+
debug_info += f"تعداد سطرها: {len(df)}\n"
|
| 146 |
+
debug_info += f"ستونها: {list(df.columns)}\n\n"
|
| 147 |
+
|
| 148 |
+
# بررسی چند سطر اول
|
| 149 |
+
for i in range(min(3, len(df))):
|
| 150 |
+
ref_text = str(df.iloc[i]['Reference_text'])
|
| 151 |
+
anon_text = str(df.iloc[i]['anonymized_text'])
|
| 152 |
+
|
| 153 |
+
debug_info += self.debug_text_analysis(ref_text, anon_text, i)
|
| 154 |
+
|
| 155 |
+
print(debug_info) # نمایش در console
|
| 156 |
+
|
| 157 |
# محاسبه متریکها برای هر سطر
|
| 158 |
precisions = []
|
| 159 |
recalls = []
|
| 160 |
f1_scores = []
|
| 161 |
|
| 162 |
+
total_entities_found = 0 # شمارنده کل موجودیتهای یافت شده
|
| 163 |
+
|
| 164 |
for index, row in df.iterrows():
|
| 165 |
precision, recall, f1 = self.evaluate_single_row(
|
| 166 |
row['Reference_text'],
|
|
|
|
| 170 |
precisions.append(round(precision, 4))
|
| 171 |
recalls.append(round(recall, 4))
|
| 172 |
f1_scores.append(round(f1, 4))
|
| 173 |
+
|
| 174 |
+
# شمارش موجودیتها برای debugging
|
| 175 |
+
ref_entities = self.extract_entities_from_text(str(row['Reference_text']))
|
| 176 |
+
pred_entities = self.extract_entities_from_text(str(row['anonymized_text']))
|
| 177 |
+
total_entities_found += sum(len(entities) for entities in ref_entities.values())
|
| 178 |
+
total_entities_found += sum(len(entities) for entities in pred_entities.values())
|
| 179 |
|
| 180 |
# اضافه کردن ستونهای جدید
|
| 181 |
df['Precision'] = precisions
|
|
|
|
| 185 |
# ذخیره نتایج
|
| 186 |
self.results_df = df
|
| 187 |
|
| 188 |
+
# پیام وضعیت شامل اطلاعات debugging
|
| 189 |
+
status_message = f"ارزیابی انجام شد. کل موجودیتهای یافت شده: {total_entities_found}"
|
| 190 |
+
if total_entities_found == 0:
|
| 191 |
+
status_message += "\n⚠️ هیچ موجودیتی تشخیص داده نشد! لطفاً فرمت دادهها را بررسی کنید."
|
| 192 |
+
|
| 193 |
+
return True, status_message, df
|
| 194 |
|
| 195 |
except Exception as e:
|
| 196 |
return False, f"خطا در پردازش فایل: {str(e)}", pd.DataFrame()
|