NetworkedData
One of the important ideas behind the concept of the SemanticWeb is "Networked Data", aka LinkedData
That is, data that exists in pieces, spread out over the Internet, that hooks up with other data, and can be combined together.
In a certain sense, then, the WikiPedia:Domain_Name_System is a part of the "Semantic Web."
The Domain Name System is great, but it only works for the Domain Name System, and programs that know how to recognize its data.
Because networked data is valuable, there are efforts to make it easier to make data formats that are networked.
RDF is a data layer that developers can put their own data format on top of, and automatically get Networked Data.
For example, suppose you were going to make a format for describing video game animation frames. You could write that data format on top of RDF. If someone ever wanted to use some animation frames from another data source across the Internet, they could make use of the rdfs:seeAlso property. When your game ran and started loading the animation cells, the RDF interpretation libraries would see the rdfs:seeAlso property, and follow it up for the additional information that was needed. After finding the necessary data, it would likely be cached locally.
Lists, dictionaries (or "hashes"), numbers, and strings are common parts of many data structures in our scripting languages. Formats like RDF add "see also" to that basic vocabulary of data structures.
It is basically a late-binding link: "If it turns out you need this information, I'll tell you where it is, and we can automatically pull it in for you."
Discussion
(none yet!)