This project is a collaboration between machine learning researchers and social scientists. It is a joint effort by a group of undergraduates, graduate students, postdocs, and professors. Meet our team members here! Check our medium posts for further reading here!
We aim to use AI techniques for building a more trustworthy social media. We work on online trolling detection, public discussion monitoring, social network analysis, and so on. Our research asks: what a healthier public communication environment would be like in this era? And how can machine learning help?
Here are some featured projects:
Understanding the prominent #metoo movement: What are people talking about and how are different topics discussed?
A word-embedding approach to discover new keywords for dynamic data collection with applications on multiple topics and discussion space:
Finding Social Media Trolls: Dynamic Keyword Selection Methods for Rapidly-Evolving Online Debates. In Neurips workshp: AI for Social Good, 2019. Paper
Tracking Social Media Movements with Dynamic Keyword Algorithm. In American Political Science Association Annal Meeting and Exhibition, 2020.
Dynamic Algorithms for Social Medial Troll Detection. In ICML WIML Un-Workshop, 2020.
A topic modeling method for capturing topic distributions and keywords that reveals how #MeToo got viral in 2017: Understanding the Evolution of the #MeToo Movement Over Time Using Topic Models. In ICML WIML Un-Workshop, 2020. Presentation
A technical report on how to collect data for online discussions monitoring: Reliable and Efficient Long-Term Social Media Monitoring. Under review.
An analysis of leadership and communication dynamics between politic figures using social media data and NLP approaches: Legislative Communication and Power: Measuring Leadership from Social Media Data