Effective Document Classification using Novel Enhanced Long Short Term Memory-based Namib Beetle Optimization Algorithm

Abstract

The process of classifying documents involves grouping them into predetermined groups according to their content. Automating document processing, increasing efficiency, and enhancing decision-making in a variety of enterprises as well as organizations are all made possible by document categorization. This technique has been widely used for text categorization applications such as sentiment analysis, topic modelling, and spam filtering. Further research is required to develop more complex techniques as well as algorithms that will increase the accuracy as well as efficacy of intelligent document categorization. Therefore, this research paper accomplishes the document classification using the novel intelligent deep learning methodology. Initially, the data is gathered from the online sources that includes a group of documents as well as their respective categories. The collected data is subjected to the pre-processing that is done with the help of tokenization, normalization, removal of header/footer, removal of stop words, and stemming of words. The output from the pre-processing is given to the feature extraction phase for extracting the features using the Chi-square approach. The extracted features enter the final classification phase, which is accomplished using the novel Enhanced Long Short Term Memory (ELSTM), in which the parameter tweaking of LSTM is performed by the nature inspired optimization algorithm called Namib Beetle Optimization (NBO) algorithm. The main objective function behind the entire novel document classification methodology is the maximization of accuracy. Finally, this innovative ELSTM-NBO classifies the output into various categories such as image processing, deep learning, data mining, sports, networks, and machine learning respectively. All things considered, the innovative ELSTM-NBO technique for document classification represents a practical as well as effective way to manage and organize massive amounts of textual data. The proposed ELSTM-NBO for the document classification model in terms of accuracy is 11.39%, 22.17%, 6.25%, and 3.03% better than SSW-based SVM, DoCA, MHGAT, and HCMBO respectively. Similarly, the proposed ELSTM-NBO for the document classification model in terms of F1 Score is 24.99%, 16.43%, 2.98%, and 2.86% advanced than SSW-based SVM, DoCA, MHGAT, and HCMBO respectively.