Why do we need Vector Databases? Why not just use relational database?

in the last video we tried and understood what are embeddings what are vectors in this particular video we try and understand what are vector databases and why do we need them as the name might indicate Vector databases are databases which help us store and manage vectors efficiently right as the name indicates uh why do we need Vector databases uh let's try and understand that with a simple example imagine you're trying to organize a giant music collection right uh a regular database might help you store information like song title artist and genre this is great if you want to find a specific song by name but not as helpful if you're trying to look for similar songs to the one that you like so here's where Vector databases come in instead of just storing text it can actually capture the music's Essence using vectors now with this Vector database you can search for songs that sound similar to the ones that you like even if they have different titles artists albums it doesn't matter Vector databases can help you find other songs which are like in the same neighborhood regular databases uh are like filing cabinets uh for music sheets right they they help you find specific details but Vector databases can act like recommendation systems suggesting you new music as per you as per what you already like so uh in this particular example the only thing that's different is uh this is a rag application the only thing that's different is instead of Facebook uh Vector DB we are using chroma Vector DB why are we using chroma because chroma allows us to store the database on disk now to import chroma I've used Lang chore chroma if I show you here I first installed it and then I've imported it here in code right and the last thing that is different is instead of uh loading the document first and then chunking it I've I've used a document loader API that is provided by langore community so what we get is uh the pages the the chunks of that particular document and the embeddings and we pass that into chroma do from documents what it will do is it will create a vector database using the text that I've provided using the embeddings function embeddings object that I've created and it will persist it into my chroma DB directory because I've named it that way right so that's stored in db2 but next what I'm doing is for my actual similarity search if you see I've used db3 so what I'm doing is is I'm retrieving that particular database from disk so you can store it and and open it up later that's that's how it will work so I've loaded it up up again in a different object I ran similarity search I ran QA chain and I'm getting my response uh what are the skills these are the skills for this particular candidate let me show you what chrom chrom ADB looks like on database so there are two files in total the first one being this bin file which stores the actual vectors uh in the database but this file is unreadable what we can though read is this sqlite file this will store the actual uh database tables indexes how they are stored right if you can see some part of it is readable where it creates tables it create indexes so this part is readable to us and that's how CHR RB stores the vectors so that was it about what are vector databases and why do we need them uh thank you for watching see you in the next one