Functional requirements for historical and interval extensions to the relational model

摘要

In recent years a lot of divergent approaches have been proposed for the management of Valid Time (Historical) data. However, no systematic effort has been reported, concerning the identification of the properties of such a model. In the present work, it is initially shown that all proposed Valid Time (VT) models can be applied to areas of practical interest, not related to VT data management. All VT models are also classified with respect to two orthogonal parameters, the way time is represented, and the level at which it is incorporated (tuple or attribute). This enables to identify that two reference VT models can be specified, VT-1NF, a simple extension to the conventional relational model and VT-NESTED, a more general one, which supports, in addition, relation-valued attributes. The properties of these models are identified. Two more reference interval relational models are proposed, I-1NF and I-NESTED, which support any type of interval data. I-NESTED is the most general, in that it can be applied to all the areas in which all others are applicable. Results are also reported, concerning the evaluation of all VT models.