Imagine I have a personal collection of recipes I like to cook, and the recipes link to foods with relation ingredient. When my friend comes over for dinner, I could ask them what foods in my kitchen they like, and instantly query all recipes containing those foods. Maybe I’ve added how long each recipe takes to cook, which appears as one of the many RDF triples I can use to filter and sort the matching recipes.
Knowledge graphs enable incredibly rich, structured queries. This is the promise and power of semantic web. Otherwise all you can do is rudimentary keyword searching.
The vision of the semantic web is to enable these rich queries over the entire collection of human knowledge, not just personal notes.
I see this exact problem with bookmarking tools. I have a large collection of bookmarks with many kinds of tags (topical, meta-tags like !important, !soon, !followup, !share, ...), etc. But the reality is those tags are not sufficient to search for what I need - adding a piece of content and searching for it are far removed in time (weeks, months, sometimes years), and "future me" doesn't remember the context something he put in with. The today-me can't anticipate which of the pieces of content will be useful and to what end.
So I don't know _what to search for_ or _how to search for it_ == but just as importantly, I don't know _what use to label something for_ and _what terms/structure_ to add to it.
I think the semantic web must have assumed we'll tag/label things correctly, and the reality is we don't do it because it's on average too hard.
RDF models facts/knowledge as subject-predicate-object statements called triples. Each part of the triple is called a resource and has a URL.
You wouldn’t bookmark pages, you would state facts about them in RDF using a vocabulary of resources.
A hypothetical semantic note taking app would let you insert many facts (RDF triples) about something you find relevant into your own knowledge graph, and the facts you insert would be constructed from your chosen vocabularies (https://www.w3.org/standards/semanticweb/ontology).
Roam/bookmarks/hypertext is the Stone Age compared to RDF knowledge graphs. https://en.m.wikipedia.org/wiki/Resource_Description_Framewo...
My current hypothesis is that [[wikilinks]] on top of the block concept popularized by tools like [[roam]] (lists in Markdown) are enough to implement a small lightweight DSL that maps to [[rdf]]. See [[agora actions]] for an early stage example representative of the direction I want to go in.
Ubuntu/Gnome's builtin search engine (Tracker) for god knows what reason uses RDF/SPARQL and anyone who wants can run "tracker sparql <query>" from the command line or their app/library/tool to query all kinds of things on their system.
Its been around for a long long time but count how many people/projects use it.