Abstract

Using deep learning models like Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Deep Neural Networks (DNN), and Random Multimodel Deep Learning (RMDL), this review paper provides an extensive examination of Document Information Retrieval (DIR). The study carefully analyses and assesses the accuracy of these individual models and ensemble strategies in DIR tasks, drawing on a close reading of more than 25 research papers.