Studying the Effects of Emotion and Stereotyping on Social Media Popularity with Deep Learning
Keywords
No Thumbnail Available
Authors
Issue Date
2018-08-06
Language
en
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Social media have become major platforms for societal discourse. Therefore, we intend to understand their dynamics. In in the current study, we investigate social media popularity using the circumplex model of emotion. We present an automated sentiment analysis method based on natural language processing and deep learning, that dissociates between valence and arousal for social media content. Subsequently, we assess the viability of our research method to concepts beyond emotion with a third variable, namely stereotyping. Based on human annotations, we classify a Twitter corpus on the aforementioned variables with deep learning algorithms. Our algorithms perform well in general, and are to be used for future reference. With negative binomial regression analysis, we report that valence, arousal, and stereotyping all significantly predict content popularity. We conclude that content sentiment should be assessed on multiple dimensions, and that machine learning and hypothesis testing combined are suitable for social media studies and social science in general. Therefore, we present future research topics and make our full method and data available to other researchers.
Description
Citation
Supervisor
Faculty
Faculteit der Sociale Wetenschappen