Sarcasm Relation to Time: Sarcasm Detection with Temporal Features and Deep Learning
摘要
This paper presents a deep learning-based framework to detect sarcasm in relation to time. Deep N-gram features generated using the FastText algorithm, combined with temporal handcrafted temporal features are used to train several machine learning classifiers. Experimental results show that Logistic Regression performs the best among all the classifiers. The introduction of the handcrafted temporal features has also whosn to improve overall detection performance when compared to existing works in the field.