Brian O'Grady
👤 SpeakerAppearances Over Time
Podcast Appearances
So now when I'm doing my ranking, there's this kind of latent association between...
disjoint words right like burger is associated with mcdonald's that you would not be able to capture in traditional text search but like with what's called semantic search where you're representing data has when you're representing text as vectors you're able to essentially do like if you remember from your sort of on like pre-calc days when you take like you know you drew on like a quadrant
no pun intended, you drew a quadrant, right?
You have like two points and you calculate like the literal distance between them.
And then you take like, you can do this in hyperspace.
You can have a vector that is a representation of text that you can then say, let's find literal distance between this representation of text as a vector and this other text piece of text that is also represented as a vector and calculate the literal distance between them to show what is
most relevant to the user's query.
So that's kind of like the very technical way of explaining it.
But the way I kind of try to think about it is people get really hung up on this idea of like, well, what is this guy talking about?
He's talking about text as vectors.
But like, we actually need to take a step back here and say, what is text itself?
Like what, when we say tech, we kind of take text for granted.
And what it really comes down to is if you think about it at a very granular level, text is simply a representation of language and meaning that we use to communicate with each other and to embody knowledge, right?
That's why like there's a lot of, I don't know if you ever read any of those books where it's like a book about a book that is powerful, right?
And when people read it, it like transforms their lives and everything, right?
Like there's tons of books out there that have this, where books are sort of like this
independent entity that are able to influence things around them.