Three Proposals for a Third-Generation Temporal Data Model

We present three general proposals for a nextgeneration temporal data model. Each of these proposals express a synthesis of a variety of contributions from diverse sources within temporal databases. We believe that the proposals may aid in bringing consensus to the area of temporal data models. The current plethora of diverse and incompatible temporal data models has an impeding eect on the design of a consensus temporal data model. A single data model is highly desirable, both to the temporal database community and to the database user community at large. It is our contention that the simultaneous foci on the modeling, presentation, representation, and querying of temporal data have been a major cause of the proliferation of models. We advocate instead a separation of concerns. As the next step, we propose a data model for the single, central task of temporal data modeling. In this model, tuples are stamped with bitemporal elements, i.e., sets of pairs of valid and transaction time chronons. This model has no intention of being suitable for the other tasks, where existing models may perhaps be more appropriate. However, this model does capture time-varying data in a natural way. Finally, we argue that exible support for physical deletion is needed in bitemporal databases. Physical deletion requires special attention in order not to compromise the correctness of query processing.