In the late 1990's, Tim Berners-Lee, creator of HTTP and the first web browser, started talking about his vision of the future of the web. He called this future the Semantic Web.
The vision of the Semantic Web is compelling. In this Scientific American article from 2001, for example, Berners-Lee and his co-authors tell a story about Lucy and Pete working with intelligent software agents to arrange a doctor's appointment for Lucy and Pete's mother. The agents collect and evaluate data from Lucy's, Pete's, and their mother's calendars, along with the web sites of clinics, a doctor-rating service, a mapquest-like service, etc. Every step of these hypothetical software interactions is clearly within the abilities of computers--no exceptional level of simulated intelligence is required. Thus, the prospect of software being able to help Lucy and Pete in this way is especially seductive, since it seems merely a matter of writing the software to do the job, and getting enough people to make machine-readable data ("meta-data") available to the agents. The story of Lucy and Pete has helped inspire an extensive effort to develop tools to help make it a reality.
However, the Semantic Web has been slow to take off. This may be due to the ambitiousness of the project. Coming up with easily used tools for representing arbitrary meaning in a machine-readable form is no mean feat, and it's not surprising that it would take some time. On the other hand, there are those who doubt that the premises of the Semantic Web are sound. See, for example, this particularly pointed criticism by Cory Doctorow.
Though Berners-Lee is a sharp guy, and his earlier project worked out pretty well, I find some of his writing on the Semantic Web a bit frustrating. This sidebar from the Scientific American article, for example, answers the question "what is the killer app of the Semantic Web?" with "the Semantic web". That is, he declines to offer a concrete answer to people who are struggling to get their minds around a highly abstract vision. Perhaps a better answer would have been "that's hard to say at this point, but shopping agents and distributed calendar coordination look very promising".
From my point of view, the Semantic Web project lacks a critical mass of concrete applications. There are some promising projects to be sure, but the ones I have seen lean towards the very general rather than the specific. I want to see a book-shopping agent, or an agent for aggregating and summarizing blog entries on a particular topic, or a system of agents to implement a distributed game of some kind, or system for merging astronomical measurements from a variety of observatories. Take those cool Semantic Web tools and make something that people can understand, and from which they can generalize new ideas.
If you are on this team, you will develop a Semantic Web application using the Semantic Web tools. The project will have the following phases:
The Semantic Web's W3C Homepage
Developing Semantic Web Services, by H. Peter Alesso
A Semantic Web Primer, by Grigoris Antoniou