BEGIN:VCALENDAR
VERSION:2.0
PRODID:ics.py - http://git.io/lLljaA
BEGIN:VEVENT
DESCRIPTION:Come see CS tenure-track candidate Laura Biester present methods of studying mental health based on data from social media websites! Assumptions will be challenged\, and time itself will factor into language processing in a way not often seen before.\n\nThursday • 12/1/22 • 3:00 - 4:00 PM\nTutt Science Center 221\nRated PG-13 (considerable undergraduate computer science experience recommended)\n\nFull abstract: \nIncreased computational power and the presence of large\, user-generated datasets have stimulated advancements in artificial intelligence in the past decade. Now\, it has become feasible to study aspects of human behavior using observational data from the internet. In this talk\, I will present methods that I have developed to study individual and population-level mental health through data from social media websites. These methods build upon techniques in machine learning and natural language processing\, supervised machine classification\, time series analysis\, and topic modeling. The first project that I will present quantifies the effects of the COVID-19 pandemic on post frequency and language use on social media support forums. The second project questions the assumption that more data leads to more robust models by considering temporal context in determining whether a person is depressed. Together\, these works lead to interesting findings about human behavior\, and show how and why time should be considered as an important variable in natural language processing and computational social science.\n\nhttps://today.coloradocollege.edu/events/5710
DTEND:20221201T230000Z
LOCATION:Tutt Science Center 221
DTSTART:20221201T220000Z
SUMMARY:Talk: Computational Linguistic Models of Mental Health!
UID:952f5a71-2d85-4992-bcef-7fc0de155669
END:VEVENT
END:VCALENDAR