Text mining social media to explore concepts and sentiments associated with mental health topics or issues
Method Stream: Text Mining
ECs: Only 14 EC (standard, no new/own data collection. Applicable in case of a clinical internship)
Description:
The ubiquity of social media potentially provides rich opportunities for research to explore real-world opinions and attitudes towards various topics. Topics related to mental health are frequently discussed on social media such as Reddit. In this project you will apply text-mining analysis (e.g., topic modeling and sentiment analysis) to explore concepts and sentiments associated with specific (e.g., positive) mental health topics or issues.
You are free in selecting a specific topic of your interest related to mental health. Previous theses have for instance explored postings on narcissism, ADHD in women, pro-anorexia posts and post-traumatic growth. You can scrape textual data yourself from media such as Reddit or specific discussion forums.
The overarching goal of the project is to explore if and what we can learn from using big-data text mining techniques of social media data about salient topics and people’s sentiments associated with mental health (issues). If relevant, subsequent analyses could focus for instance on changes in sentiment over time, differences in sentiments between topics or subreddits, or associations with number of likes or comments.
Text-mining analyses can be done with the free (point-and-click) Orange Data Mining software, or with R or Python if you want to (learn to) do some more advanced methods like transformer-based natural language processing.