Today the internet and especially social networks are the source of numerous up-to-date information. These include not very helpful information like “Dear drivers drive carefully. I’m on the way with my bicycle.”, but also useful information like “The closure of the motorway A14 was lifted, currently only one lane per direction”. To be able to use these information, e.g. to verify or extend the information from official sources, it have to be filtered and analyzed. As a result of the overwhelming message stream, this can only happen automatically. By the example of traffic information the aim of the project “AUSWEG” is to extract information from unstructured sources on the basis of domain-specific guidelines. Along with the use of semantic technologies and dialog based machine learning, also the generation of active queries to the authors of the messages shall be examined.
