Comparative Study of Deep Learning Models for Document Classification

In the last few years, deep learning has shown promising improvements in natural language processing (NLP). LSTM, Bi-LSTM, and Convolutional Neural Networks (CNNs) are examples of neural networks (NNs) that are utilized for a variety of NLP applications, including sentiment analysis, information retrieval, and document classification. We compare these models to the BBC News dataset consisting of five categories: business, entertainment, politics, sport, and technology in this paper. Each model is vectorized using TF-IDF, Word2vec, and Glove. And then the performance of each model is evaluated by precision, recall, and F1-Score. In comparison with other models, Word2vec and CNN achieved a higher F1-Score.