COVID News Articles (2020 - 2022)

The dataset encapsulates approximately half a million news articles collected over a period of 2 years during the Coronavirus pandemic onset and surge. It consists of 3 columns - title, content and category. title refers to the headline of the news article. content refers to the article in itself and category denotes the overall context of the news article at a high level. The dataset encapsulates approximately half a million news articles collected over a period of 2 years during the Coronavirus pandemic onset and surge. It consists of 3 columns - title, content and category. title refers to the headline of the news article. content refers to the article in itself and category denotes the overall context of the news article at a high level.

This dataset can be used to pre-train large language models (LLMs) and demonstrate NLP downstream tasks like binary/multi-class text classification. The dataset can be used to study the difference in behaviors of language models when there is a shift in data. For e.g., the classic transformers based BERT model was trained before the COVID era. By training a masked language model (MLM) using this dataset, we can try to differentiate the behaviors of the original BERT model vs the newly trained models.