How can our personal data improve search result?

Sometimes I want to read an article that I read before, but it’s difficult for search engine to know which artiche I’ve read.
Sometimes I want to read some articles that I’ve never read before, but search engine (or feed of reddit or quera) still recommend something I already know about.

If my pod can store some meta knowledge about what knowledge I may have, then recommendation and search result will be more accurate, and save my time of filtering info.

So I hope there can be an APP that extract ontology such as topic and events from the webpage I’ve read, and build up a personal knowledge graph for me. (I know this is very hard)

But I’m not sure, how to represent that “meta knowledge”. And after we have personal knowledge graph, how can it improve our search result and feed? What should webpage owners change?

Below is something I stumbled across a while ago which got me thinking. It refers to a paper about using personal metadata to customise the browser experience, but would fit your use case very well too.

The paper is linked in the following post (on the SAFE Network forum). It opened my eyes to the power of RDF to provide enhanced personal web services without pooling of data in centralised cloud services owned by Google, Amazon etc.