Sochiatrist: Automatically Predicting Emotion From Social Messaging Data
Abstract
Major Goals: The proposed research addresses the following research questions: 1) How accurately can emotion measured through self-report or through biomarkers be predicted purely from social messaging data? In what conditions can specific emotions or biological responses be more predictable, especially as an indicator of social support? 2) Which aspects of social messaging are predictors of emotion, and in what way do different social features about the frequency, content, timing, and recipients of messages influence affect? 3) Are predictions better or worse for different populations, particularly those who have been admitted for mental health concerns? Inversely, if we know a mental health condition of a person, does that change how we may predict their emotional states? However, the contributions of this work go beyond answering these research questions. The proposed work goes through the full cycle of developing a prediction model, reporting the results from multiple study populations, and releasing the software for social scientists. We will publish lessons from developing the prediction model, including both the statistical factors of the social messaging features, and the comparison between different prediction algorithms for emotional states. Perhaps most useful for the social science community, the PIs research group will release the Sochiatrist software that performs automated extraction of social messaging data for social scientists in multiple research groups across the country, who then use that data to make inferences about their participants. The software will be offered along with documentation and support for those social scientists to use without needing additional support from our team. Social messaging data can be sampled cost-effectively from a large number of participants.
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 31, 2023
- Accession Number
- AD1229494
Entities
People
- Jeff Huang
Organizations
- Brown University