A Comparative Analysis for Effective Text Document Classification Using Machine Learning Algorithms and Deep Convolution Neural Network

Abstract

The enormous amount of text documents keeps on increasing day by day to a greater extent on the web. Almost 80% of the data are available in the form of text on the web. The voluminous of text documents in this digital era requires organizing them consistently which facilitates the information retrieval process. Hence text mining plays a vital role in the process of information retrieval. This paper is focusing on text document classification that has its wider applications in information retrieval, document indexing based on controlled vocabulary, word sense disambiguation, generating hierarchical categorization of web pages, spam detection, email categorization, sentiment analysis, named entity recognition(NER), topic labeling, web search and ranking, document summarization etc. Text document classification belongs to the category of Natural Language Processing tasks where the machine itself automatically categorizes the text documents based on the content to its classes. A lot of manual effort and time is saved by using automatic text document classification. Text document consists of a huge, sparse, non-uniform distribution of features. Mining informative features and performing text classification still exist as a challenging task. This paper contributes techniques involved in text document classification and performs comparative analysis by using machine learning algorithms and deep learning algorithms. The proposed model is experimented with 20-Newsgroups dataset and also evaluated using different performance measures. It has been proven that the proposed model using deep convolution neural network gives superior performance when compared to machine learning algorithms. It gives accuracy 96.3% precision 100%, recall 100% and f1-score99.8%.