The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence) What NLP Tells Us About COVID-19 and Mental Health with Johannes Eichstaedt - #400
Aug 13, 2020
In this conversation with Johannes Eichstaedt, an Assistant Professor of Psychology at Stanford University, they explore how he blends physics and psychology to analyze mental health trends using big data. Johannes discusses the fascinating use of Twitter data to uncover psychological impacts during COVID-19, revealing insights into societal behavior changes and mental health fluctuations. He also highlights the challenges of capturing nuanced language variations across different communities, shedding light on the dynamic nature of social norms during the pandemic.
AI Snips
Chapters
Transcript
Episode notes
Advantages of Social Media Data
- Social media offers unobtrusive data collection, unlike surveys.
- This allows for ecological assessment as people behave naturally in digital spaces.
Social Desirability Bias
- Social desirability bias on social media leads to suppression of negative emotions and failures.
- This doesn't invalidate data, but requires larger datasets for accurate analysis.
Avoiding Twitterology
- Avoid "Twitterology", focus on studying people, not Twitter itself.
- Discard Twitter-specific signals (retweets, replies) to improve signal-to-noise ratio.