A Probabilistic Correlative Convolutional Extreme Learning Model for Semantic Classification of Social Media Documents

Abstract

Text document classification is a fundamental task in Natural Language Processing (NLP) that involves dispersing predefined categories or labels to a given text. It is widely used for organizing, filtering, and analyzing content. In order to address these issues, a novel method called Probabilistic Bivariate Correlative Convolutional Extreme Learning Machine (PBiC-CELM) for efficient document classification with higher accuracy and minimum time consumption. The Convolutional Extreme Deep Belief Network is a hybrid deep learning architecture that integrates the strengths of Convolutional Neural Networks (CNNs) and Extreme Learning Machine (ELM) for advanced feature learning and text document classification. PBiC-CELM method includes three different processes. First, numbers of text documents are collected from dataset. These text documents are given as input to hybrid architecture CELM. The text preprocessing is carried out in convolutional layer where tokenization, Stop Word Removal and Word stemming are performed. The pre-processed text is transmitted to hidden layer 2. In that layer, Least Angle Regression Analysis is carried out for efficient keyword extraction from the pre-processed texts. Then, the extracted keywords are transmitted to ELM architecture for document classification. In that layer, Bivariate Correlation Coefficient is employed for classifying the document with extracted keywords. In this way, an accurate semantic document classification is performed with higher accuracy and minimal time consumption. The analyzed results demonstrate that proposed PBiC-CELM method achieves efficient performance outcomes, including higher accuracy, precision, recall, F1-score, and specificity with minimized time consumption and error rate when compared to existing methods and state-of-art-methods.

Keywords: Semantic text document classification, extreme learning machine, Convolutional Neural Networks (CNNs), Least Angle Regression results, Bivariate Correlation Coefficient.