Linear Repeating Points

Abstract

Temporal databases have been a research subject for many years, though one might wonder why so much attention has been devoted to this topic; indeed, time can easily be introduced as additional attributes in the relational model. Of course, specific operations need to be performed with temporal attributes, such as dealing with date formats, but such operations are not really problematic, and have long been available in commercial systems. Nevertheless, the feeling remains that time is special. Clearly, introducing time in a database requires some thought: Is one going to deal with valid time, transaction time, or both; is the basic unit a point or an interval; is time continuous or discrete; if discrete, what is the basic granularity? Nevertheless, once these questions have been answered, a direct representation by additional attributes appears at first to be adequate. With respect to querying, however, the need to express constraints on the time information arises in many cases, for instance in a query such as “which projects have overlapping schedules?” Moreover, when looking at the evaluation of queries, one notices that specific operations, such as compacting overlapping intervals, are necessary but are not directly supported in relational query languages.