Semantic Search

Definitions

1. Search based on the meaning of queries rather than keyword matching. Converts both queries and documents into vector embeddings and finds the nearest matches in the embedding space. 'How to fix a broken pipe' finds plumbing AND Unix content based on context. Powered by transformer models and vector databases.

   In plain English: Search that understands what you mean, not just the exact words you type.

   Example: "Semantic search finds relevant docs even when users use different terminology — searching 'container orchestration' matches an article titled 'Managing Docker at Scale.'"

Related Terms

• LLM
• Vector Database
• Token
• Vector Database
• Attention
• Approximate Nearest Neighbor
• Beam Search
• BERT
• Byte Pair Encoding
• CLIP
• Contrastive Learning
• Contrastive Loss
• Dense Retrieval
• ELMo
• Knowledge Graph Embedding
• Learning to Rank
• Matryoshka Embeddings
• Slot Filling
• CLIP Model
• Neural Search