Semantic Search and Neural Information Retrieval
summary
Covers building a production neural search pipeline that replaces brittle term expansion with dense embeddings and nearest-neighbor lookup, demonstrated by indexing 2,000 Netgear Nighthawk reviews on the Zero Semantics platform to surface context like “social time wasting sites” and cross-language queries. Reports a 27% relative jump in F1 when swapping LSTM style models for transformers, describes vector database (annoy, hnsw) based retrieval, snippet extraction, re-ranking and the need for score calibration.
AI-written summary · about
chapters (7)
- 0:00Introduction to neural search
- 1:56Limitations of keyword search
- 7:23Neural search demonstration
- 10:07Historical AI breakthroughs
- 18:57Modern search architecture
- 23:39Comparing commercial solutions
- 25:21Real-world decision support
related talks
25mSemantic Search
32mBudget Friendly Semantic Search With Neural Sparse Search
46mNatural Language & Search, Improve your Search w/ Vector capabilities
47mThe Neural Search Frontier #bbuzz
13mA Beginner’s Guide to Semantic Search
42mResearch in the Age of AI
32mBuilding semantic search using Deep Learning
26mAdvancing Multi-Modal Search Capabilities in Search Pipeline