The PEL project seeks to recognize complex event in videos, here in the context of basketball and volleyball games. In our approach we argue that holistic reasoning about time intervals of events, and their temporal constraints is critical to overcome the noise inherent to low-level video representations. Thus we use probabilistic event logic (PEL) for representing efficiently temporal constraints among events. Specifically, our approach leverages the spanning-interval data structure for compactly representing and manipulating entire sets of time intervals without enumerating them. In this project, we derive a MAP inference algorithm for PEL that addresses the scalability issue of reasoning about an enormous number of time intervals and their constraints in a typical video.