Automatic Humour Detection in Tweets using Soft Computing Paradigms

Humour has been an integral entity of human life with its lingual, social and psychological aspects. Due to its vast applications and increasing popularity on social media platforms, enabling computers to process humour has become very crucial. Therefore, we intent to aid task of computational analysis of humour by employing deep learning and machine learning techniques for humour detection. Social media websites such as Facebook, Twitter, Reddit etc are an abundant source of humorous content. Hence, we use micro-blogging website Twitter as the focus of this study. We created a symmetrically distributed dataset containing 4000 funny and plain tweets and applied seven supervised classification algorithms to perform the task of detecting which tweets contain humour. Accuracy, recall and precision have been used as the metrics of performance.