Text Classification Accuracy Enhancement Using Deep Neural Networks

Abstract

Machine-learning systems are part of the field of artificial intelligence that automatically learn models from data to make better decisions. Natural language processing (NLP), using object and learning approaches, performs well in statistical tasks, such as text classification or sentiment mining. Currently, in the field of natural language processing (NLP), various models, methods, and architectures have been proposed and a wide range of NLP tasks have been developed. Most natural language processing (NLP) problems can be formulated as classification problems (decide on the class of this object depending on the object and its context). In this field, machine learning methods and especially using neural network models are the most used. In this research, we propose a deep learning architecture for text classification that can influence model performance. The basic steps of this proposed method include text pre-processing (including sentence segmentation, punctuation, rooting, speech tagging, and vector conversion), text feature extraction, and LSTM algorithm-based classification model construction. The dataset used in this study is 20NewsGroup and IMDB. The evaluation results are provided in the form of MSE, Precision, Recall, Accuracy, and F-Measure evaluation criteria.