Abstract: Annotation of corpora is a labor-intensive and time and resources consuming task. Active annotation is an active learning based semiautomatic annotation procedure. The goal of Active Learning is to speed-up and make easier the human annotation process. In Active Annotation we use the models learnt during the annotation process in order to find annotation errors and cases that are hard to be automatically annotated with the features used by the learner. The analysis of these case alows to add new features to the set and optimize the use of previously used features.

Presentation Slides