Temporal expression extraction with extensive feature type selection and a posteriori label adjustment

The proposed approach ranked 3rd in the TempEval-3 challenge (task A) as the best performing machine learning-based system among 21 participants.