Category:
Category:
Semantic Search
Category:
Retrieval & RAG
Definition
Search based on meaning rather than keywords.
Explanation
Semantic search retrieves documents based on their meaning using embeddings and vector similarity. Unlike keyword search, it identifies relevant content even when phrasing differs. Semantic search is foundational for RAG systems, enterprise knowledge assistants, and agent memory. High-quality embeddings and chunking determine its accuracy.
Technical Architecture
Query → Embedding → Vector DB → Similarity Search → Reranking → LLM
Core Component
Embedding model, vector store, similarity metric, metadata filtering
Use Cases
Knowledge assistants, document QA, enterprise search, RAG
Pitfalls
Noisy results if embeddings are poor; missing exact-match numeric or acronym retrieval
LLM Keywords
Semantic Search, Vector Search, Embedding Search
Related Concepts
Related Frameworks
• RAG
• Embeddings
• Reranking
• Hybrid Retrieval
• Semantic Search Architecture Map