Semantic search

0:09 okay welcome to the session for generative generative AI for beginners uh my name is Dave Glover I'm A Cloud Advocate on Bas in Sydney Australia and uh this is for lesson six building search applications Al righty so what we're going to cover in the session we're going to look at what semantic search is you might a term you might have heard banded around we're going to talk about vectors and embeddings again a term that you might have heard but just kind of what the difference is between them we're going to talk about Vector stores and we're going to cover demos so the goal of this session so ultimately at the end of the session you're going to have a better idea how vectors play a crucial role in enabling effective semantic search so hopefully that's what you're going to pick up out of the session now when I first uh started talking and learning about

0:56 generative AI um one of the terms that was banded around a lot was semantic search and I must admit at the time I didn't really know what semantic search was but a bit of digging around and I kind of got the general idea of how this works and hopefully this will help you as well so semantic search is really kind of intent based search or boiling down to what are the The Core Concepts you're thinking about you want to look for so take an example if you talk about my dream watch now if you did a keyword search you would end up with results which were about dreams and which about watches which is probably what you weren't what not what you were looking to find now if you were doing semantic search what that would do is it will boil down to the concept probably of my ideal watches which is what you're really thinking about and then You' find things about ideal watches and you'd probably expand on that search term so the key thing about semantic surge is around intent or around the concept that you're looking for as opposed to to keyword search which would find dreams and watches now semantic search is kind of like a really pivotal point or

2:07 pivotal concept when it comes to building large language model applications and you'll often find these applications take a big dependency on what we call semantic search and embeddings okay so you've heard about the term vectors no doubt you're learning about those in University of school and um you might have heard the word embeddings you might be thinking