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.