Supercharged Search with Semantic Search and Vector Embeddings
summary
Demonstrates building semantic search by generating 384‑dim and 1,124‑dim embeddings with OpenAI or local Olama via Microsoft.Extensions.AI, storing them in SQL Server 2025 vector type and querying with EF Core, then adding a DiskANN index for ANN. Uses 50,000 news rows; in‑memory kNN is ~100 ms, plain DB scan ~300 ms, and indexed ANN drops below ~50 ms while keeping ~95% recall. Code and demos show end‑to‑end generation, storage, and measured query times.
AI-written summary · about
chapters (12)
- 0:00Semantic search and embeddings
- 1:05Lexical vs semantic search
- 3:16Semantic search with LLMs
- 8:08Storing vector embeddings
- 17:53Vector databases and SQL Server
- 23:13Demo: Embeddings for news
- 26:29In-memory semantic search
- 30:08SQL Server vectors with C#
- 37:27Vector indexing and ANN
- 43:59Evaluating ANN recall
- 49:05Demo: Speed comparison
- 50:44Multilingual semantic search
related talks
28mUse OpenSearch as Your Vector Search Database To Optimize GenAI Accuracy!
22mVector Search Made Simple: Getting Started With OpenSearch for AI Applications
01h 01mDrop the Bass with Embedding and Vectors in Azure AI Search
55mVector Search Made Simple: Getting Started with OpenSearch for AI Applications
28mSponsored Session: AI-Powered Semantic Search with Cassandra: Today and Tomorrow
22mGenerative AI application stack and providing long term memory to LLMs
39mFrom Search to Insight: Leveraging OpenSearch for Scalable, AI-Driven Search Experiences
30mHow to Decipher User Uncertainty with GenAI and Vector Search