Detecting humorous expressions on social media

Humorous content underpins much of the material that goes “viral” online, whether across social media, on websites, or even in text messages. Despite the important part it plays in many people’s lives there is scant research that has focused on humor in social media. New work in the International Journal of Information Technology and Management looks to remedy that situation.

Suzanne Elayan, Martin Sykora, Thomas W. Jackson, and Ejovwoke Onojeharho of the Centre for Information Management at Loughborough University, UK, suggest that understanding humorous content and the dynamics it follows on social media could help improve applications that carry out sentiment analysis of online content. The team has focused on the heterogeneous nature of humor on the well-known and popular microblogging platform, Twitter. Specifically, whereas the earlier work that does exist concentrates on English language humor, the present study investigates Arabic content. In this context, the team suggests that “Automated humor detection in its own right has potential in understanding public reactions and should be explored in future studies.”

While there is growing evidence from the animal kingdom that other species might understand and react to comical situations, the notion of humor, seems at its deepest levels to be a human trait with many social, psychological, and biological benefits. It is a trait that has provided entertainment, social coherence, and release to humanity for thousands of years. We can consider jokes, anecdotes, wordplay and puns, irony, and schadenfreude/sarcasm/mockery to be the basic categories of humor. In the era of social media, we now have at our fingertips a way to share a joke with millions, if not billions of people almost instantaneously.

Of course, everyone’s sense of humor differs and there are some people who may find a particular piece of humor to be side-splittingly hilarious while another might perceive the same joke as unfunny, or worse, highly offensive and upsetting. It is the subjective nature and the many unknown factors surrounding humor and its effects that have perhaps meant that little solid research has been undertaken that gives us solid results about the topic. The team’s review of the literature nevertheless points to a way to develop tools that might be able to extract humorous sentiment from given content in an automated manner that would be useful to researchers in sociology and psychology as well as more broadly the study of social media itself.

“It is hoped that this paper will aid in establishing further research activities in this exciting and interesting field, helping to lead this new stream of research,” the team concludes.