This paper addresses the problem of entity alignment in attribute-rich event-driven graphs. Unlike many other entity alignment problems, we are interested in aligning entities based on the similarity of their actions, i.e., entities that participate in similar events are more likely to be the same. We model the generative process of this problem as a Bayesian model and derive our proposed algorithm from the posterior predictive distribution. We apply our Hierarchical Entity AlignmenT (HEAT) algorithm to two datasets, one on publications and the other on financial transactions, derived from real data and provided to us by an external collaborator.