An intelligent model based on integrated inverse document frequency and multinomial Naive Bayes for current affairs news categorisation

Digital technologies, their product and services have empowered the masses to generate information at a faster pace. Digital technologies based information sharing platforms such as news websites and social media platforms such as Facebook, Twitter, Instagram, What's app etc have flooded the information space due to the easy generation of information and dissemination to the masses instantly. Information classification has been an important task, especially in newspapers and media organisations. In another area also, information or text classification has an important role to play so that important and vital information can be classified based on the already predefined categories. In journalism, editors and resources persons were allocated the task to recognise and classify the news stories so that they can be placed in the predefined categories of economy and business news, political news, social news, editorial section, education and career, and sports information etc. Nowadays the process of classification and segregation of textual information has become challenging due to the flow of diverse, vast information. Additionally, the pace of information and its updates, access and competition among the media House have made it more challenging. Hence automated and intelligent tools which can classify the information and text accurately and efficiently is needed to reduces human efforts, time and increase productivity. This paper presents an intelligent, efficient and robust intelligent machine learning model based on Multinomial Naive Bayes(MNB) to classify the current affairs news stories. The proposed Inverse Document Frequency(IDF) integrated MNB model achieves classification accuracy of 87.22 per cent. The experiment results are also compared with other machine learning models such as Logistics Regression(LR), Support Vector Machine(SVM), K-Nearest Neighbours(KNN) and Random forest(RF). The results demonstrate that the presented model is better in term of accuracy and may be deployed in real world information classification and media domain to improve the productivity, efficiency of the current affairs news classification process.