This document discusses research into modeling the temporal formation and evolution of online communities. It proposes: 1) Modeling each user's interests over time as a topic space, building a weighted graph between users based on topic similarity over time, and using graph clustering to identify communities of like-minded users. 2) Analyzing causality between community behaviors over time to better understand influence and predict future behaviors, noting challenges include establishing temporal precedence and improving predictions using causal information. The research aims to better understand how interests drive similar temporal behaviors between users to form online communities.