Automatic detection of ironic expressions based on the Implicit Display Theory

It is not easy for a computer to automatically detect irony. Because in order to understand the irony in a sentence, it is necessary to infer the meaning hidden behind the word from the branch. Common sense and common understanding of reality are required as background knowledge for detecting irony. The literal meaning and the meaning intended by a speaker do not often match, and this makes it a difficult task in natural language processing. In this research, the effectiveness of automatic detection of irony based on the Implicit Display Theory was verified. The Implicit Display Theory was proposed in 1997 as a comprehensive irony theory, and its validity and superiority have been shown as a theory representing irony research. We selected ironic sentences for which the Implicit Display Theory holds from a large-scale ironic corpus, and performed two-class classification using the deep learning model to evaluate the application of implicit display theory to natural language processing.