"So, computer, please find me all documents that contain research information about a drug that can cure cancer, developed anywhere in the world" - this is a classic question we would like to ask a computer. Actually, its so classic that it is defined as an example in the 1992 version in the TREC test data.

The DFKI Spin-Off SemVox may provide something that helps realizing this. They are combining ontologies with speech interaction:
The SemVox technology enables the user to employ various applications without having to resort to traditional operating concepts such as keyboards or remote controls. Using our technology the user is free to choose between a number of modalities such as speech, gestures, keyboard or mouse or a combination thereof.



Their technology incorporates a heterogenous set of modules that can be remixed to allow different application scenarios. Part of their demos is to tell the computer to "find me an action film". Nice side-effect: using the speech-synthesis module offered by SVOX, the computer will talk back to you (press release in german).

So - this is a next step to the semantic web,as Vint Cerf has put it:
I’m almost certain you’ll see products emerging that will allow you to orally interact with the network
Sure, it is nearly here, and you can buy the tools for it off-the-shelf. And I guess SemVox is open for investors :-)

What is really funny, is that today we are very close to actually answer the questions defined as scientific goals in 1992 (for example here, page 64, I was not able to find the original TREC-1 set).

I have seen the SemVox system live at CeBit, I was demoing NEPOMUK (and advertising my gnowsis.com startup) 5 meters away from them and we had great fun demoing our products to each other. Here is a picture of Jan, one of the founders: