DeepHumor: a novel deep learning framework for humor detection

The automation of humor detection is a complex task due to the semantic structure of the textual content. In this paper, an automatic humor detection model named as DeepHumor is proposed. The proposed model is based on the combination of convolutional neural network (CNN) and long short term memory (LSTM). The highway network is also incorporated in the proposed model to enhance the performance. The hybrid model uses CNN layers for feature extraction with LSTM layers for sequence learning. To overcome the overfitting problem in the proposed model, dropout layers are added. The performance of the proposed DeepHumor model is compared with seven recently developed techniques over Yelp user review dataset. The proposed DeepHumor model attained the significantly better performance than the exiting techniques in terms of precision, recall, accuracy and F1-measure.