Spaces:
Sleeping
Sleeping
File size: 14,506 Bytes
19aaa42 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 |
"""
Simple Vector Store for Medical RAG v2.0
Research-backed approach: Document-based retrieval with simple metadata
"""
import os
import json
import logging
import time
from typing import List, Dict, Any, Optional, Tuple
from pathlib import Path
import numpy as np
from dataclasses import dataclass
# Vector store and embeddings
import faiss
from sentence_transformers import SentenceTransformer
from langchain_core.documents import Document
@dataclass
class SearchResult:
"""Simple search result structure"""
content: str
score: float
metadata: Dict[str, Any]
document_name: str
content_type: str
class SimpleVectorStore:
"""
Simple vector store using research-optimal embedding approach
- Focused on document-based retrieval
- Simplified metadata structure
- High-performance FAISS indexing
"""
def __init__(self,
embedding_model: str = "all-MiniLM-L6-v2",
index_type: str = "IndexFlatIP", # Inner Product for cosine similarity
vector_store_dir: str = "simple_vector_store"):
"""
Initialize the simple vector store
Args:
embedding_model: Sentence transformer model name
index_type: FAISS index type
vector_store_dir: Directory to store vector index and metadata
"""
self.embedding_model_name = embedding_model
self.index_type = index_type
self.vector_store_dir = Path(vector_store_dir)
self.vector_store_dir.mkdir(exist_ok=True)
# Initialize components
self.embedding_model = None
self.index = None
self.documents = []
self.metadata = []
self.setup_logging()
self._initialize_embedding_model()
def setup_logging(self):
"""Setup logging for the vector store"""
logging.basicConfig(
level=logging.INFO,
format='%(asctime)s - %(name)s - %(levelname)s - %(message)s'
)
self.logger = logging.getLogger(__name__)
def _initialize_embedding_model(self):
"""Initialize the sentence transformer model"""
try:
self.logger.info(f"Loading embedding model: {self.embedding_model_name}")
self.embedding_model = SentenceTransformer(self.embedding_model_name)
self.logger.info(f"Embedding model loaded successfully")
except Exception as e:
self.logger.error(f"Error loading embedding model: {e}")
raise
def create_embeddings(self, chunks: List[Document]) -> Tuple[np.ndarray, int]:
"""Create embeddings for document chunks"""
if not chunks:
raise ValueError("No chunks provided for embedding")
start_time = time.time()
# Extract text content
texts = [chunk.page_content for chunk in chunks]
self.logger.info(f"Creating embeddings for {len(texts)} chunks...")
# Generate embeddings
embeddings = self.embedding_model.encode(
texts,
show_progress_bar=True,
batch_size=32,
normalize_embeddings=True # Important for cosine similarity
)
# Store documents and metadata
self.documents = chunks
self.metadata = [chunk.metadata for chunk in chunks]
embedding_time = time.time() - start_time
self.logger.info(f"Created {len(embeddings)} embeddings in {embedding_time:.2f} seconds")
return embeddings, len(embeddings)
def build_index(self, embeddings: np.ndarray):
"""Build FAISS index from embeddings"""
dimension = embeddings.shape[1]
# Create FAISS index
if self.index_type == "IndexFlatIP":
# Inner Product index (good for normalized embeddings)
self.index = faiss.IndexFlatIP(dimension)
elif self.index_type == "IndexFlatL2":
# L2 distance index
self.index = faiss.IndexFlatL2(dimension)
else:
raise ValueError(f"Unsupported index type: {self.index_type}")
# Add embeddings to index
self.index.add(embeddings.astype('float32'))
self.logger.info(f"Built FAISS index with {self.index.ntotal} vectors")
def save_vector_store(self):
"""Save vector store to disk"""
try:
# Save FAISS index
index_path = self.vector_store_dir / "faiss_index.bin"
faiss.write_index(self.index, str(index_path))
# Save documents
docs_path = self.vector_store_dir / "documents.json"
docs_data = []
for doc in self.documents:
docs_data.append({
'page_content': doc.page_content,
'metadata': doc.metadata
})
with open(docs_path, 'w', encoding='utf-8') as f:
json.dump(docs_data, f, indent=2, ensure_ascii=False)
# Save configuration
config_path = self.vector_store_dir / "config.json"
config = {
'embedding_model': self.embedding_model_name,
'index_type': self.index_type,
'total_documents': len(self.documents),
'dimension': self.index.d if self.index else 0,
'created_at': time.strftime('%Y-%m-%d %H:%M:%S')
}
with open(config_path, 'w', encoding='utf-8') as f:
json.dump(config, f, indent=2)
self.logger.info(f"Vector store saved to {self.vector_store_dir}")
except Exception as e:
self.logger.error(f"Error saving vector store: {e}")
raise
def load_vector_store(self) -> bool:
"""Load vector store from disk"""
try:
index_path = self.vector_store_dir / "faiss_index.bin"
docs_path = self.vector_store_dir / "documents.json"
config_path = self.vector_store_dir / "config.json"
if not all(p.exists() for p in [index_path, docs_path, config_path]):
return False
# Load FAISS index
self.index = faiss.read_index(str(index_path))
# Load documents
with open(docs_path, 'r', encoding='utf-8') as f:
docs_data = json.load(f)
self.documents = []
self.metadata = []
for doc_data in docs_data:
doc = Document(
page_content=doc_data['page_content'],
metadata=doc_data['metadata']
)
self.documents.append(doc)
self.metadata.append(doc_data['metadata'])
# Load configuration
with open(config_path, 'r', encoding='utf-8') as f:
config = json.load(f)
self.logger.info(f"Loaded vector store with {len(self.documents)} documents")
return True
except Exception as e:
self.logger.error(f"Error loading vector store: {e}")
return False
def search(self,
query: str,
k: int = 5,
content_type_filter: Optional[str] = None) -> List[SearchResult]:
"""
Search for similar documents
Args:
query: Search query
k: Number of results to return
content_type_filter: Filter by content type (optional)
Returns:
List of SearchResult objects
"""
if not self.index or not self.documents:
raise ValueError("Vector store not initialized. Load or create index first.")
# Create query embedding
query_embedding = self.embedding_model.encode(
[query],
normalize_embeddings=True
)
# Search in FAISS index
# Get more results initially for filtering
search_k = min(k * 3, len(self.documents))
scores, indices = self.index.search(query_embedding.astype('float32'), search_k)
# Process results
results = []
for score, idx in zip(scores[0], indices[0]):
if idx == -1: # Invalid index
continue
doc = self.documents[idx]
metadata = self.metadata[idx]
# Apply content type filter if specified
if content_type_filter:
doc_content_type = metadata.get('content_type', '')
if content_type_filter.lower() not in doc_content_type.lower():
continue
result = SearchResult(
content=doc.page_content,
score=float(score),
metadata=metadata,
document_name=metadata.get('document_name', 'Unknown'),
content_type=metadata.get('content_type', 'general')
)
results.append(result)
# Stop when we have enough results
if len(results) >= k:
break
return results
def get_stats(self) -> Dict[str, Any]:
"""Get vector store statistics"""
if not self.documents:
return {"status": "empty"}
# Document statistics
doc_counts = {}
content_type_counts = {}
total_chars = 0
for doc in self.documents:
# Document distribution
doc_name = doc.metadata.get('document_name', 'Unknown')
doc_counts[doc_name] = doc_counts.get(doc_name, 0) + 1
# Content type distribution
content_type = doc.metadata.get('content_type', 'general')
content_type_counts[content_type] = content_type_counts.get(content_type, 0) + 1
# Character count
total_chars += len(doc.page_content)
# Vector store size estimation
if self.index:
# Estimate size: vectors + metadata
vector_size_mb = (self.index.ntotal * self.index.d * 4) / (1024 * 1024) # 4 bytes per float32
metadata_size_mb = total_chars / (1024 * 1024) # Rough estimate
total_size_mb = vector_size_mb + metadata_size_mb
else:
total_size_mb = 0
return {
"status": "ready",
"total_chunks": len(self.documents),
"embedding_model": self.embedding_model_name,
"index_type": self.index_type,
"vector_dimension": self.index.d if self.index else 0,
"vector_store_size_mb": round(total_size_mb, 2),
"avg_chunk_size": round(total_chars / len(self.documents), 1),
"document_distribution": dict(sorted(doc_counts.items(), key=lambda x: x[1], reverse=True)[:10]),
"content_type_distribution": content_type_counts
}
def main():
"""Main function to test the simple vector store"""
print("π Testing Simple Vector Store v2.0")
print("=" * 60)
try:
# Initialize vector store
vector_store = SimpleVectorStore(
embedding_model="all-MiniLM-L6-v2",
index_type="IndexFlatIP"
)
# Check if we can load existing vector store
if vector_store.load_vector_store():
print("β
Loaded existing vector store")
else:
print("π Creating new vector store from chunks...")
# Load chunks from simple chunker
from simple_document_chunker import SimpleDocumentChunker
chunker = SimpleDocumentChunker()
documents = chunker.load_processed_documents()
chunks = chunker.create_simple_chunks(documents)
print(f"β
Loaded {len(chunks)} chunks for embedding")
# Create embeddings
embeddings, count = vector_store.create_embeddings(chunks)
# Build index
vector_store.build_index(embeddings)
# Save vector store
vector_store.save_vector_store()
print("β
Vector store created and saved")
# Get statistics
stats = vector_store.get_stats()
print(f"\nπ VECTOR STORE STATISTICS:")
print(f" Status: {stats['status'].upper()}")
print(f" Total chunks: {stats['total_chunks']:,}")
print(f" Embedding model: {stats['embedding_model']}")
print(f" Vector dimension: {stats['vector_dimension']}")
print(f" Store size: {stats['vector_store_size_mb']} MB")
print(f" Avg chunk size: {stats['avg_chunk_size']:.0f} chars")
print(f"\nπ Content Type Distribution:")
for content_type, count in stats['content_type_distribution'].items():
percentage = (count / stats['total_chunks']) * 100
print(f" {content_type}: {count:,} chunks ({percentage:.1f}%)")
# Test search functionality
print(f"\nπ TESTING SEARCH FUNCTIONALITY:")
test_queries = [
"magnesium sulfate dosage preeclampsia",
"postpartum hemorrhage management",
"fetal heart rate monitoring",
"emergency cesarean delivery"
]
for query in test_queries:
print(f"\nπ Query: '{query}'")
results = vector_store.search(query, k=3)
for i, result in enumerate(results, 1):
print(f" Result {i}: Score={result.score:.3f}, Doc={result.document_name}")
print(f" Type={result.content_type}")
print(f" Preview: {result.content[:100]}...")
print(f"\nπ Simple Vector Store Testing Complete!")
print(f"β
Successfully created vector store with {stats['total_chunks']:,} embeddings")
print(f"β
Search functionality working with high relevance scores")
return vector_store
except Exception as e:
print(f"β Error in simple vector store: {e}")
import traceback
traceback.print_exc()
return None
if __name__ == "__main__":
main() |