Query languages for a heterogeneous temporal database

Abstract

Time is an important dimension in databases. In a conventional database, out-of-ate information is usually replaced by current information to keep the database up-to-date. In many applications, discarding old information is inappropriate. A temporal database incorporates the notion of time. Objects in a temporal database are not deleted. On the contrary, they are retained and time-stamped to indicate their periods of existence in the real world. Over the last few years, a number of temporal relational database models have been proposed. This thesis reviews these models and proposes a heterogeneous model. We also develop a relational algebra and a tuple calculus for this model and prove their equivalence. The model and query languages capture the concept of "always" and "sometime" in natural language. In addition, they provide powerful operations with respect to temporal properties of information contained in tuples. When compared to existing approaches, fewer operations are needed to express complex queries. As a result, fewer intermediate scratch pads will be used during query execution. Consequently the space and time complexity introduced by the temporal dimension in temporal databases will be reduced.