I am sort of looking into building categorisation system which will basically ingest continuous streams of data through various pipelines in real time for eg, Twitter posts and the comments in those posts, data from website sources in forms of articles.

I want the system to organise the data in topics and subtopics for eg:

Topic — Laptops

Subtopic — Acer reveals brand new XYZ laptop

I don’t intent to use generative AI to write subtopics, I am more looking forward to form sentance from the data sources like for eg if Acer launches XYZ laptop, then all the Twitter posts and articles would include something along the lines of “Acer revealed/launched blah blah blah”.

I just need a direction from where to start because I am kinda lost how could implement it

I never even got past building data streams. Let alone setting up the model to do categorisation