Presenter's Name(s)

SandhyaFollow

Primary Faculty Mentor Name

Christopher M. Danforth

Status

Graduate

Student College

College of Engineering and Mathematical Sciences

Program/Major

Computer Science

Second College (optional)

College of Engineering and Mathematical Sciences

Second Program (optional)

Complex Systems

Primary Research Category

Health Sciences

Secondary Research Category

Social Sciences

Presentation Title

Word Embeddings to quantify Depressive Language in Twitter

Time

1:00 PM

Location

Silver Maple Ballroom - Health Sciences

Abstract

How do people discuss mental health on social media? Can we train an algorithm to recognize differences between discussions of depression and other topics? Can an algorithm predict that someone is depressed from their tweets alone? In this project, we collect tweets referencing `depression’ over a seven year period, and train word embeddings to characterize linguistic structures within the corpus. We find that neural word embeddings capture the contextual differences between “depressed” and “healthy” language. The best performing model for the prediction task is Long Short-Term Memory (LSTM) with 70% test accuracy. Finally, we train a similar model on a much smaller collection of tweets authored by individuals formally diagnosed with depression. The results suggest social media could serve as a valuable screening tool for mental health.

This document is currently not available here.

Share

COinS
 

Word Embeddings to quantify Depressive Language in Twitter

How do people discuss mental health on social media? Can we train an algorithm to recognize differences between discussions of depression and other topics? Can an algorithm predict that someone is depressed from their tweets alone? In this project, we collect tweets referencing `depression’ over a seven year period, and train word embeddings to characterize linguistic structures within the corpus. We find that neural word embeddings capture the contextual differences between “depressed” and “healthy” language. The best performing model for the prediction task is Long Short-Term Memory (LSTM) with 70% test accuracy. Finally, we train a similar model on a much smaller collection of tweets authored by individuals formally diagnosed with depression. The results suggest social media could serve as a valuable screening tool for mental health.