News Categorization using Deep Learning Models – BERT and LSTM

News Categorization using Deep Learning Models – BERT and LSTM

Ankita Thombre1, Vrishali Chakkarwar2

1M. Tech, Department of Computer Science and Engineering, Government College of Engineering, Aurangabad.

2Assistant Professor, Department of Computer Science and Engineering, Government College of Engineering, Aurangabad.

---------------------------------------------------------------------***---------------------------------------------------------------------

Abstract - In this digital era, there are various platforms that continuously generate news on various content. The news generated, range from environment like global issues to personal matters of celebrities and even that of common people. Platforms like Twitter and Instagram have facilitated in the increase of online content generation by the user, which in turn can be extracted as new trends in market, thus resulting in data essential for businesses. This chaos of news articles over the internet may lead to loss of significant data, reason being sheer amount of information and lack of enough resources to access it. Gaining insight from this chaos requires proper classification of news articles, into appropriate categories. Deep Learning methods tend to understand the structure and semantics of natural language, thus resulting in better classification of news text data. Applying transformer architecture to deep learning techniques can enhance the understanding of languages by machines. This work applies BERT and LSTM models for news text classification task.

Key Words:  Text classification, news categorization, Bi-directional Encoder Representations from Transformers (BERT), Long Short–Term Memory (LSTM), deep learning, neural networks.