Junming Huang

I am an Associate Research Scientist at Princeton University, specializing in computational social science. My research interests include examining how gender and racial identities affect scientific careers, leading to population-level disparities. I also investigate the formation and evolution of human behaviors and opinions, particularly under media and peer influence. Utilizing large-scale data, my research employs statistical methods such as machine learning, network analysis, and natural language processing. Prior to my current role, I was a Postdoctoral Research Associate at the Center for Complex Network Research at Northeastern University. I hold a PhD in Computer Science from the Chinese Academy of Sciences and a Bachelor's degree in Physics from Tsinghua University.

Science of science

The field of Science of Science is dedicated to quantitatively deciphering the universal patterns driving scientific discoveries using extensive bibliometric data, with an aim to offer insights for fostering academic progress. My research delves into disparities within academia, a realm complicated by interplaying factors such as discipline, country, and institutional prestige. By reconstructing detailed career trajectories of scientists from bibliometric data, I tackle this complexity for in-depth explanation of gender inequality and impact of scientist nationality. Those findings have been featured in PNAS and have garnered media attention from Nature Index, Associated Press, and The Wall Street Journal.

Human opinions and peer influence

Human opinions are important in social norm discussions and complicated in dynamics. In the era of explosive usage of social media and social networks, human online behaviors are influenced by close friends and opinion leaders, referred to as peer influence. The recent availability of large-scale online opinion data provides rich resources to observe and understand human opinions, from trending to scaling, from mobility to proximity, from the reaction to external shocks to the consequences on physical space behaviors. Read more

Probabilistic models

Everything has a chance. We construct probabilistic models that tackle an array of practical issues, encompassing personalized recommender systems, causal inference, virus epidemics, and survey harmonization. The toolbox for these endeavors includes Bayesian networks, probabilistic matrix/tensor factorization, time series analysis, latent variables, network analysis, among others. Read more