Opinion mining in Dutch Hansards

The question is addressed if opinion mining techniques can be successfully used to automatically retrieve political viewpoints in Dutch parliamentary publications. Two specific tasks are identified: automatically determining subjectivity in the publications and automatically determining the semantic orientation of the subjective parts. A collection of recent parliamentary publications has been collected and a golden standard annotation is created on both subjectivity and orientation. Following this, models based on subjectivity lexicons and machine learning algorithms are evaluated on automatic classification. Overall results tend to be dominated by machine learning algorithms, but methods based on subjectivity lexicons provide promising results. Based on the results it is concluded that opinion mining techniques can indeed be successfully used to automatically retrieve political viewpoints in Dutch parliamentary publications.