[3Rin4] Interactive 1

[3Rin4-34] Automatic detection of ironic exressions with an attention method

〇shunya fukuta¹, Seiki Matoba², Hirotoshi Taira¹ (1.department of Information Science and Technology, Osaka institute of technology, 2.Faculty of Information Science and Technology, Osaka Institute of Technology)

Keywords:Irony Detection, Deep Learning, Natural Language Processing

It is not easy to detect ironic expressions automatically. One of the reasons is that the surface expression of the sentence dosen't stand for the meaning literally .It is effective to consider the context for automatic detection of ironic expressions. The accuracy of automatic irony detection was improved by using BERT, which can be classified using long-range context information.As a result of the experiment, the method using BERT exceeded the method used as the baseline and showed its effectiveness.