Michael Christensen, Brigham Young University
Social and Behavioral Sciences
Suicide is a major problem for students in Utah middle and high schools. Since adolescents spend a large portion of their time on social media like Facebook and Twitter, there is a wealth of information we can learn about their personalities, moods, and interests by exploring their online interactions with others and specifically the statuses and messages they post publicly. We have created a Facebook app to mine this data and report the mood of a logged-in user’s entire network based on the individual classification of community members’ posts. We contribute to the Public Health field by aggregating suicide-risk factors and facilitating intervention, the motivation being to help others better identify and help those who are at-risk for suicide based on their online behavior. We contribute to the Computer Science field by creating a machine learning algorithm that can classify text into one of several fine-grained mood categories, learning to identify more than just positive or negative sentiment. In addition, our algorithm has the ability to update online by receiving feedback from the users on how well or poorly it classified the text of their friends’ posts. We describe our algorithm and report on preliminary results about its performance on real-world data.