Well, the title of the video is misleading. There is not much discussion about the best practices. It is an in-depth video of the AI Seach service and its SDK. Here are the timestamps of topics covered. Intro to topic and demo of vector search: 02:00 Example with index profile definition and searching: 18:00 Vector search strategies: 24:00 Hybrid search and semantic re-ranking: 26:30 AI Search indexing strategies: 33:00 Manual indexing/Bring your own embedding: 34:00 Integrated vectorization/Indexing docs in the cloud: 38:00 Skillsets for indexers: 49:00 Integrated vectorization RAG samples: 50:15 Advanced features: Analyzers, Scoring profiles: 58:00
Thanks for adding the timestamps! I'll re-title this session the next time we give it. In this case, I meant that hybrid search+ ranker is the best practice, and many devs are new to that, but there are certainly many more AI Search best practices that we don't cover.
How do you know what to index? For example if I have uploaded a number files ranges from pdf, doc etc to use as my data for ai search to look at… then what am I indexing?
Try preparing the files, separate every file by section. And then index with the different fields. I have done that adjustment of data with gpt or other LLM model.
Well, the title of the video is misleading. There is not much discussion about the best practices. It is an in-depth video of the AI Seach service and its SDK. Here are the timestamps of topics covered.
Intro to topic and demo of vector search: 02:00
Example with index profile definition and searching: 18:00
Vector search strategies: 24:00
Hybrid search and semantic re-ranking: 26:30
AI Search indexing strategies: 33:00
Manual indexing/Bring your own embedding: 34:00
Integrated vectorization/Indexing docs in the cloud: 38:00
Skillsets for indexers: 49:00
Integrated vectorization RAG samples: 50:15
Advanced features: Analyzers, Scoring profiles: 58:00
Thanks for adding the timestamps! I'll re-title this session the next time we give it. In this case, I meant that hybrid search+ ranker is the best practice, and many devs are new to that, but there are certainly many more AI Search best practices that we don't cover.
very well explained and detailed session. Thanks for sharing
How do you know what to index? For example if I have uploaded a number files ranges from pdf, doc etc to use as my data for ai search to look at… then what am I indexing?
Try preparing the files, separate every file by section. And then index with the different fields.
I have done that adjustment of data with gpt or other LLM model.
can I have the not book please?
Wow mine favourite guide @pamela fox