Data Models with Multiple Temporal Dimensions: Completing the Picture

Abstract

There is a widespread recognition that valid and transaction times are the fundamental temporal dimensions of any fact relative to a database. There are, however, temporal aspects of facts that cannot be naturally modeled by means of them. A remarkable limitation of valid and transaction times is that they do not allow one to distinguish between retroactive and delayed updates. A third temporal dimension, called event time, has been proposed in the literature, which makes it possible to model retroactive, on-time, and proactive updates.In this paper, we first refine the notion of event time by showing that one event time does not suffice to model relevant phenomena, and then we introduce a further temporal dimension, that we called availability time, which can be viewed as the information system counterpart of the real-world event time. We conclude the paper by outlining current and future work directions.