Search engines use semweb Semweb v. English Semweb as social media Semweb as Wikipedia: Epistemology Semweb as Wikipedia: Traversal
The semantic web lets you model reality in a computer.
When you ask Google the circumference of the Earth, it answers directly. It just "knows". That is because "the Earth" is an object in Google's semantic web, and "[spehrical thing] has a circumference of [distance]" is a relationship in that web, and the Earth is in such a relationship.
All information is relationships. The semantic web is a language for making relationships explicit to a computer. English, by contrast, is a language for making relationships explicit to another human.
Suppose you discover that glass is a liquid. You could write somewhere, "Glass is a liquid." Once those four words are written, anybody who reads them can infer many things about glass and liquids -- for instance, that the molecules in glass move freely with respect to each other, or that something does not have to flow fast to be a liquid.
Language lets us share information, which in turn lets us share inquiry, discovery, science. Sharing information is what makes humanity so powerful.
Computers can't read English (yet), but they can process the semantic web.
Imagine a social media platform where people could relate posts to concepts: "This post is about politics. This post is about soccer and heroism. This post is funny and fictional."
With those relationships explicit to the computer, you could construct your own news feed algorithm: "Show me politics unless it's near my bedtime. Show me so-and-so's art. Show me the things the people I love love ..."
The semantic web makes it possible to draw such relationships -- drawn by the author, by other humans, or inferred automatically. (Certain kinds of metadata, such as who wrote something, are easy to detect automatically. Other kinds, such as what a piece of writing is about, can be inferred too, at least to some extent.)
In fact, no centralized social media platform is needed for this. A news feed algorithm could crawl your favorite blogs (assuming they provide semantic metadata) and from those assemble for you what looks like a Facebook feed.
In most media, there is a sharp distinction between data and metadata. For instance, in Wikipedia you can easily see what's on Guatemala's page, but where that information came from can be hard or impossible to determine.
The semantic web, by contrast, encodes statements, and statements about statements, and statements about statements about statements, etc. in the same way. Assuming the data exists, a semweb wiki user could, with equal ease, view data like:
a statement
who wrote the statement
and why -- that is, the purpose for which they wrote it
the evidence (other statements) that imply it
the other statements it implies
who disputes the statement, and why
A semantic web wiki would be a boon to science, reporting, and politics.
Most media carries a fixed presentation. A book, or a wikipedia article, has a fixed table of contents, and a fixed order.
With linked semantic data, a user could tailor the data's presentation to what they need to know. If a statement is not obvious, ask for the evidence it comes from. If you are interested in controversy, ask for only those statements that have been disputed.
The semantic web will even let you view a topic that was never (explicitly) written about! A page about the history of cloth in India, for example, could be assembled from a query like "history of cloth & history of India", which would collect every statement that qualifies for both categories.