Automatic satire detection

We introduce the novel task of determining whether a newswire article is "true" or satirical. We experiment with SVMs, feature scaling, and a number of lexical and semantic feature types, and achieve promising results over the task.