Convolutional Neural Network Based Simile Classification System

Figurative Language is a kind of language that express ideas with a meaning different from its actual literal interpretation. Social media makes use of a wide range of figurative languages. Among them Simile, Sarcasm, Irony and humor are mostly used. Simile essentially compares two different things and often occur in tweets. People make use of indirect interpretations to express their views. Thus arises the need of detecting other figurative languages occurrence in similes. For proper understanding of text, occurrences of sarcasm, irony and humor in similes is necessary. Current approaches mainly focus on detection of one or two figurative languages at once. This paper proposes classification of different figurative languages at once using deep learning concept such as Convolutional Neural Network (CNN) and machine learning concepts like Support Vector Machine (SVM), Decision Tree, K-nearest neighbor (KNN), Gaussian Naive Bayes (GNB) classifier. Also, compared the performance of different classification models.