Spaces:
Sleeping
Sleeping
File size: 11,672 Bytes
d12790d |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 |
# # src/classifier.py
# from sentence_transformers import SentenceTransformer
# import numpy as np
# import pickle
# class ProductClassifier:
# def __init__(self, model_path="./models"):
# self.model = SentenceTransformer("all-mpnet-base-v2")
# self.embeddings = np.load(f"{model_path}/category_embeddings_mpnet.npy")
# with open(f"{model_path}/category_metadata.pkl", "rb") as f:
# self.metadata = pickle.load(f)
# def classify(self, product_data, top_k=5):
# # Implementation here
# pass
# """
# Product Classification Engine
# Loads pre-trained embeddings and performs similarity-based classification
# """
import numpy as np
import pickle
from sentence_transformers import SentenceTransformer
from sklearn.metrics.pairwise import cosine_similarity
from typing import Dict, List, Optional
import re
import logging
from .config import (
MODEL_NAME,
EMBEDDINGS_FILE,
METADATA_FILE,
AUTO_APPROVE_THRESHOLD,
QUICK_REVIEW_THRESHOLD,
BOOST_FACTOR,
MAX_BOOST,
DEFAULT_TOP_K,
PRODUCT_KEYWORDS,
)
# Set up logging
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)
class ProductClassifier:
"""
ML-powered product classifier for insurance categorization
"""
def __init__(self):
"""Initialize classifier by loading model and embeddings"""
logger.info("Initializing Product Classifier...")
# Load the embedding model
logger.info(f"Loading model: {MODEL_NAME}")
self.model = SentenceTransformer(MODEL_NAME)
logger.info(
f"β
Model loaded (dimension: {self.model.get_sentence_embedding_dimension()})"
)
# Load pre-computed category embeddings
logger.info(f"Loading category embeddings from {EMBEDDINGS_FILE}")
self.embeddings = np.load(EMBEDDINGS_FILE)
logger.info(f"β
Loaded {self.embeddings.shape[0]:,} category embeddings")
# Load category metadata
logger.info(f"Loading metadata from {METADATA_FILE}")
with open(METADATA_FILE, "rb") as f:
self.metadata = pickle.load(f)
logger.info(f"β
Metadata loaded")
# Cache for processed texts
self.embedding_texts = self.metadata.get("embedding_texts", [])
logger.info("π Classifier ready!")
def preprocess_product(self, product_data: Dict) -> str:
"""
Preprocess product data into searchable text
Args:
product_data: Dictionary with product fields
- title (str): Product title
- product_type (str, optional): Product type/category
- vendor (str, optional): Brand/vendor name
- tags (list/str, optional): Product tags
- description (str, optional): Product description
Returns:
Processed text string for embedding
"""
parts = []
# Extract fields in priority order
title = product_data.get("title", "")
product_type = product_data.get("product_type", "")
vendor = product_data.get("vendor", "")
description = product_data.get("description", "")
tags = product_data.get("tags", [])
# 1. Title (most important)
if title:
parts.append(title)
# 2. Product type (category hint)
if product_type:
parts.append(f"Product type: {product_type}")
# 3. Brand/Vendor
if vendor:
parts.append(f"Brand: {vendor}")
# 4. Tags (keywords)
if tags:
tag_text = " ".join(tags) if isinstance(tags, list) else tags
parts.append(f"Keywords: {tag_text}")
# 5. Description (limited to 100 chars)
if description:
desc_short = description[:100].strip()
parts.append(desc_short)
return ". ".join(parts)
def extract_keywords(self, text: str) -> List[str]:
"""
Extract important keywords from product text
Args:
text: Product text
Returns:
List of detected keywords
"""
text_lower = text.lower()
found_keywords = [kw for kw in PRODUCT_KEYWORDS if kw in text_lower]
return found_keywords
def classify(
self, product_data: Dict, top_k: int = DEFAULT_TOP_K, use_boost: bool = True
) -> Dict:
"""
Classify a product into insurance categories
Args:
product_data: Product information dictionary
top_k: Number of top matches to return
use_boost: Whether to apply keyword boosting
Returns:
Classification results with confidence scores and recommendations
"""
# Preprocess product text
product_text = self.preprocess_product(product_data)
# Generate embedding for product
product_embedding = self.model.encode([product_text], normalize_embeddings=True)
# Calculate semantic similarities
semantic_scores = cosine_similarity(product_embedding, self.embeddings)[0]
# Apply keyword boosting if enabled
if use_boost:
product_keywords = self.extract_keywords(product_text)
boosted_scores = self._apply_keyword_boost(
semantic_scores, product_keywords
)
else:
boosted_scores = semantic_scores
# Get top K indices
top_indices = boosted_scores.argsort()[-top_k:][::-1]
# Format results
results = []
for rank, idx in enumerate(top_indices, 1):
category_data = {
"rank": rank,
"category_id": self.metadata["category_ids"][idx],
"category_path": self.metadata["category_paths"][idx],
"semantic_score": float(semantic_scores[idx]),
"final_score": float(boosted_scores[idx]),
"confidence_percentage": round(float(boosted_scores[idx]) * 100, 2),
}
# Add boost information if used
if use_boost:
category_data["boost_applied"] = round(
(boosted_scores[idx] - semantic_scores[idx]) * 100, 2
)
results.append(category_data)
# Determine action based on top score
top_confidence = results[0]["final_score"]
if top_confidence >= AUTO_APPROVE_THRESHOLD:
action = "AUTO_APPROVE"
reason = f"High confidence ({results[0]['confidence_percentage']}%)"
elif top_confidence >= QUICK_REVIEW_THRESHOLD:
action = "QUICK_REVIEW"
reason = f"Medium confidence ({results[0]['confidence_percentage']}%) - verify category"
else:
action = "MANUAL_CATEGORIZATION"
reason = f"Low confidence ({results[0]['confidence_percentage']}%) - needs expert review"
return {
"product_id": product_data.get("id", "unknown"),
"product_text": product_text,
"action": action,
"reason": reason,
"top_category": results[0]["category_path"],
"top_confidence": results[0]["confidence_percentage"],
"alternatives": results[1:3] if len(results) > 1 else [],
"all_results": results,
}
def _apply_keyword_boost(
self, scores: np.ndarray, product_keywords: List[str]
) -> np.ndarray:
"""
Apply keyword-based score boosting
Args:
scores: Original semantic similarity scores
product_keywords: List of keywords found in product
Returns:
Boosted scores
"""
boosted_scores = scores.copy()
if not product_keywords:
return boosted_scores
# Boost categories that contain product keywords
for idx, cat_text in enumerate(self.embedding_texts):
cat_text_lower = cat_text.lower()
matches = sum(1 for kw in product_keywords if kw in cat_text_lower)
if matches > 0:
# Boost proportional to keyword matches
boost = min(matches * BOOST_FACTOR, MAX_BOOST)
boosted_scores[idx] = min(boosted_scores[idx] + boost, 1.0)
return boosted_scores
def classify_batch(
self, products: List[Dict], top_k: int = DEFAULT_TOP_K
) -> List[Dict]:
"""
Classify multiple products at once
Args:
products: List of product data dictionaries
top_k: Number of top matches per product
Returns:
List of classification results
"""
logger.info(f"Classifying batch of {len(products)} products...")
results = []
for i, product in enumerate(products, 1):
try:
result = self.classify(product, top_k=top_k)
# Convert all numpy types to Python native types for JSON serialization
result = self._convert_to_json_serializable(result)
results.append(result)
if i % 100 == 0:
logger.info(f" Processed {i}/{len(products)} products")
except Exception as e:
logger.error(f" Error classifying product {i}: {e}")
results.append(
{
"product_id": product.get("id", f"product_{i}"),
"action": "ERROR",
"reason": str(e),
"top_category": None,
"top_confidence": 0.0,
}
)
logger.info(f"β
Batch classification complete!")
return results
def _convert_to_json_serializable(self, obj):
"""
Recursively convert numpy types to Python native types
"""
import numpy as np
if isinstance(obj, dict):
return {
key: self._convert_to_json_serializable(value)
for key, value in obj.items()
}
elif isinstance(obj, list):
return [self._convert_to_json_serializable(item) for item in obj]
elif isinstance(obj, (np.integer, np.int64, np.int32)):
return int(obj)
elif isinstance(obj, (np.floating, np.float64, np.float32)):
return float(obj)
elif isinstance(obj, np.ndarray):
return obj.tolist()
else:
return obj
# Test the classifier if run directly
if __name__ == "__main__":
print("Testing Product Classifier...")
print("=" * 80)
# Initialize classifier
classifier = ProductClassifier()
# Test product
test_product = {
"id": "test_001",
"title": "Apple iPhone 15 Pro Max",
"product_type": "Smartphone",
"vendor": "Apple Inc",
"tags": ["electronics", "mobile", "phone", "smartphone"],
"description": "Latest flagship smartphone with titanium design",
}
print("\nπ± Test Product:")
print(f" {test_product['title']}")
# Classify
result = classifier.classify(test_product)
print(f"\nπ― Classification Result:")
print(f" Action: {result['action']}")
print(f" Top Category: {result['top_category']}")
print(f" Confidence: {result['top_confidence']}%")
print(f" Reason: {result['reason']}")
print("\nπ Top 3 Alternatives:")
for alt in result["alternatives"][:3]:
print(
f" {alt['rank']}. {alt['category_path']} ({alt['confidence_percentage']}%)"
)
print("\n" + "=" * 80)
print("β
Classifier test complete!")
|