In the research project Brain-Bots, a man-machine interface is to be developed in which the communication is based on natural language. With this the existing learning application "Virtual Microbiology Laboratory" (VML) shall be extended by interaction. VML was developed at Beuth University Berlin (Beuth HS) in cooperation with an industrial partner as part of a teaching research project and is in use there. Currently, the learner1 within VML is accompanied by a virtual lab assistant (avatar). This praises or blames the learners, depending on whether they act incorrectly or correctly in the virtual laboratory. A direct response of the assistant is currently not possible. The learners can therefore not ask the avatar any technical questions about the course of an experiment, such as "What chemical do I need now?", "Why is this done?" Or "What's the best way to do it?".

For the extension of the learning application to such interaction possibilities, two software components, a language analysis system and a dialogue system as well as a knowledge base are to be developed. The language analysis system analyzes learners' requests semantically and forwards the result to the dialogue system. That chooses context-sensitive, according to the progress within an experiment, the appropriate answer based on a knowledge base. In the learning application, this answer is spoken by the virtual lab assistant via a text-to-speech engine (already widely available) and displayed in a speech bubble. The overall system has to work in real time because the learners expect constant response times from the "brain-bot" who looks after them in the lab's virtual microbiology lab. The software components created in the project should be reusable independently of each other for other applications or new experiments (in the virtual laboratory).

Projectpartners: Dr. Felix Gers (BHT, FB 6), Dr. Christian Herta (HTW, FB 4), Dr. Steffen Prowe (BHT, FB 5), Villa Hirschberg Online GmbH (Berlin)