Judging the Psychological Impact of News Articles on Rehab Patients: A Deep Learning Approach

Sentiment detection and analysis from review comments of products, blogs and other social media communication is a common practice. Lot of research work have been proposed and established in the domain. While the above may prove to be effective for sales and marketing challenges but the same effort for the sentiment analysis of news articles is scarce. But surprisingly, this can definitely establish as a potentially efficient instrument for medical domain judging the suitability of a news article for individuals with disturbed health condition like depression, heart diseases or mental retardation having risk of committing suicides. To address the same, this article presents a novel approach harnessing the potential of Machine Learning (ML) and Deep Learning (DL) to detect the sentiment of a news article so as to measure its impact on the individual readers. The proposed work uses various supervised ML algorithms such as SVM, Naïve Bayes’, Logistic Regression, Random Forest and Ensemble Method for assessing the inherent sentiment of news articles. Additionally, DL based Transformer model specifically BERT has also been employed to enhance the performance. Obtained results with various performance metrics clearly depict our model’s excellent performance.