Computational Linguistics and Information Processing

Assessment of Suicidality in Social Media Interaction

Approximately 25 million American adults will have an episode of major depression this year, and suicide is the third leading cause of death for people aged 10-14 and second among people between people age 15 - 34.  But clinicians lack measures with good predictive validity for assessing suicidality using in-person evaluations, much less effective means for assessment using online evidence. In this project we have developed a new dataset for research on suicidality assessment using anonymous data from Reddit, and we are investigating methods for obtaining reliable clinical expert judgments from online postings, scaling up human assessment using crowdsourcing, and developing computational predictive models using a range of methods including deep learning, transfer learning, and multi-task learning.