Statistic Supported Cooperative Creation of Training Corpora for the Extraction of Traffic Information from Microblogs

Microblogs like Twitter 1 are sources for a lot of useful information. But these information are unstructured. To automatically process it, these information have to be extracted with the help of natural language processing techniques. Corresponding systems for information extraction have to be trained and validated with manual annotated data. This paper presents an approach for the cooperative creation of annotated corpora for training and validation of information extraction systems supported by statistical analyses.