Study on Impact of RNN, CNN and HAN in Text Classification

Text classification is one of the major research areas in the field of Natural Language Processing. This is the task of determining the category of the text based on its content. This paper analyzes how the semantic and contextual information of the text data is encoded with different deep learning models such as Recurrent Neural Network (RNN) with Long Short Term Memory (LSTM), Convolutional Neural Network (CNN) and Hierarchical Attention Network (HAN). Each of the three models exhibits specific characteristic while representing the contextual information like RNN keeps the sequential structure of the text data, CNN captures the n-gram feature of the document, and HAN maintains the hierarchical structure of the document by considering the critical words and sentences in a document. These models were implemented on the data set, BBC News text and the results are analyzed on the basis of the various parameters like precision, recall, accuracy, F1-Score, confusion matrix and loss.