Mainstream pornography has a long and well-documented history of relying on stereotypes to deliver its viewers sexual gratification. Concerns about this have been compounded in recent years by the massive spread of online porn and the fact that it draws many young people who get to witness these stereotypes which are not rarely blatantly misogynist and racist.
Yet, despite ample accounts of these stereotypes harming the mental health and physical well-being of individuals and communities, we still know very little about why so many people from diverse social backgrounds enjoy watching stereotypes in online pornography and how we can avert detrimental effects on young people’s healthy sexual development.
The aim of this project is to break new scientific ground by using the latest tools from machine learning to trace developments in the content and popularity of gendered, ethnic and racial stereotypes in online pornography.
Specifically, it will use a web ‘scraper’ to collect visual and textual data on millions of videos from Pornhub and XVideos (the two most-visited porn websites globally and in the Netherlands) over a 15-year period. Then Structural Topic Modeling will be used to uncover latent topics within this large complex dataset and to couple these with metadata on popularity and time of posting.
Findings from this advanced, massive scale, longitudinal study, will not only be applicable to future research on the popularity and impact of pornographic stereotypes, but on the social sciences as a whole, providing new tools for analyzing large and complex (video-based) datasets.