A Deep Learning Approach for Bengali News Headline Categorization

Abstract

A huge amount of information and data is now available in seconds through the internet. Due to the availability of internet in the world, as the amount of online news is increasing on the one hand, people are interested in reading news from online news portals i.e., Facebook, Twitter, WhatsApp, Telegram, Instagram, blogs etc. Along with the increase in the amount of digital data in news portals, the number of readers is also increasing, thus the need for data classification for digital data is also increasing day by day. There are various methods of data classification, such as machine learning, deep learning, etc., as well as various data mining algorithms. Data is classified using these algorithms, so that people can make sense of the news just by reading the news headlines. Natural language processing methods are used to classify data in any language for such problems. In this paper, Bangla news is classified into 6 categories using deep learning algorithm. The categories are International, National, Sports, Entertainment, Politics and IT. In deep learning, BiLSTM and GRU algorithms have been used to classify. BiLSTM has an accuracy of 83.42% and GRU has an accuracy of 80.01%. BiLSTM has been found to have the highest accuracy among deep learning algorithms.