Abstract

ABSTRACT The rapid growth of online media and communities has increased and accelerated the spread of unverified and false information (“fake news”), with significant political, economic, and social impacts, leading the European Commission to promulgate a “Code of Practice on Disinformation.” Identifying and countering such false information is time- and labor-intensive, and could benefit from the development of tools that automatically identify and flag such information. This study explores the use of deep learning techniques to detect fake news, using decreases in the incidence of emotional vocabulary and subjectivity to enhance detection accuracy, and examines potential correlations between the emotional sentiment of news content and the movement of stock price indexes. Empirical results show that deep learning techniques can be used to effectively detect fake news, with multiple trainings effectively improving detection accuracy and reducing the loss rate. In addition, increased objectivity and the use of fewer words with high emotional sentiment increases news credibility. Finally, news sentiment was found to be correlated with the movement of three of five stock indexes examined.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call