Process analysis relies on views on event data composed of contexts that group events, e.g., as induced by a case notion, and a relation between events, such as ‘directly-follows’. Together, they cluster and structure events into classical traces or, recently, into object-centric executions or actor routines. Contexts are selected to cater to a specific analysis question. Yet, when dealing with a new process or a new analysis question, an analyst has to explore different context definitions. Materializing all resulting views results in unnecessary overhead, not only in computation and storage, but also in analysis complexity and data redundancy. In this paper, we study the construction of views on event data in an exploratory setting by introducing the notion of a context in event data and its induced view on the data. Based thereon, we define the problem of view materialization to find contexts that adequately represent the data from different perspectives. These views provide an analyst an overview of the data and enable exploratory analysis beyond preconceived case notions. We instantiate the problem for object-centric contexts and address it drawing on strategies for subset selection. We show the feasibility using experiments with synthetic and real-world event data.