Machines that get the joke - science2innovation

Although there's quite some research done on irony detection, there's still not enough data generated to use for the actual training. The solution can be a large-scale irony dataset, which will allow to create complex training models and provide more accurate sentiment analysis. In order to address the lack of irony data the study analyzes more than 2 million tweets and proposes a novel model to transfer non-ironic sentences to ironic sentences in an unsupervised way. This is the first research that generates ironic sentences to achieve a high irony accuracy with well-preserved sentiment and content.