I think Drupal 7 stands a chance of being a killer semantic app, it s the next version of a CMS with a sizeable install base and when the release is final and everyone starts upgrading suddenly tons of sites will be automatically exposing RDF in the form of embedded RDFa without any intervention on the part of users.

From a business viewpoint I think things linked data stuff like GoodRelations are a potential killer app, for example see these Scott Brinker and Priyank Mohan blog posts which discusses how BestBuy experienced a 30% increase in traffic after starting embedding GoodRelations based linked data as RDFa in their web pages.

In general linked data is an excellent way to augment your site with machine readable data and shows remarkable results in the visibility of your site, I ve recently seen a couple of presentations by guys from the BBC whose wildlife finder application (a linked data application using BBC natural history resources) already outperforms some Wikipedia pages for animals in Google search results

In biology, the interest in RDF & related technology is very high. People want both less parsing and custom code to integrate data, and more advanced queries. RDF is already providing the former; for example, UniProtKB, one of the biggest biomedical resources, are offering their data in RDF. For advanced querying we re not quite there yet, as performance and availability of good RDF data and OWL ontologies is still a bit lacking. But it s starting, check out BioGateway for an example of what can be done.

In that light, the fruits of these technologies are not individual killer apps, but all the accumulated time savings for researchers who can explore the data instead of writing ad hoc scripts and setting up temporary SQL databases and all this plumbing.

Just discovered DBpedia. It s a promising attempt to represent a lot of data from Wikipedia as RDF. You can download the entire extraction as a 16GB file, but it also has a keyword search page. Being represented as RDF allows for very specific "semantic" queries. This shows a few example queries, such as how you would find a list of the official websites of companies with more than 50000 employees. You can even query "the cloud" remotely with any wrapper that can interface with its public API, such as this Python module.

The semantic web relies on content providers taking the effort to properly annotate everything to make it machine-readable.

That is too much work for most people that are not librarians.

The real killer app seems to be something that can derive semantics from unstructured content without special markup. Look at the amazing job Google for example has done with its search-engine.

In order for a system to work, it must not place a heavy burden on its users.

A middle ground are tagging systems like here on Stackoverflow. They work pretty well, even though the tags are completely adhoc and in parts inconsistent.

Deep semantic support is interesting for a number of places where powerful queries are needed. An example was in a project I was working on recently where the service that was used for looking up where to dispatch workload to was semantically-based. RDF/SPARQL itself is interesting because it gives you pretty rich querying right off the bat, but when you add an OWL ontology in it gets even better as it means that you can answer richer queries (i.e., that ask questions closer to what the user – and their employer – really wants) while allowing the service providers to express what they re offering more clearly too. Which isn t to say that it means everyone telling everyone everything, not at all. Instead, we had parties describing what services were provided and not what configuration they were using to provision those services. And it was all empowered by the use of semantic technologies pervasively throughout the information systems.

Currently I m working on Taverna which is now using RDF to provide a souped-up logging system; in particular, users (especially scientists) can search through the wealth of information recorded much more easily than if they had to just grep through a huge text file. After all, it would be a little absurd if you had to use text mining to find out what really happened in your text mining workflow...

I was impressed with AceWiki. It s one of several wiki projects out there that s attempting to semantically organize the information people enter.

It s still a work in progress, and has some limitations. For example, it only allows people to enter sentences that conform to a subset of English vocabulary and grammar.

However, I hope these take off. There s such a huge amount of free data out there (e.g. Wikipedia) and we can t create programs to reason over it because the content is too noisy.

Protege is good for developing ontologies.