Indexical databases

Abstract

The modeling capability of today's DBMS's will need to be expanded if they are to serve the needs of tomorrow computing problems. Numerous research proposals have appeared to extend this functionality for a wider array of application areas. Among these proposals have been a plethora of suggestions for historical databases, rollback databases, and bitemporal databases.

We have described the Indexical Database Model (IDM) as a generalization of the work done in these and related areas. In this chapter we have provided an overview of the structures of this model and the operators in its algebra, and shown how one of these models — the historical relational data model (HRDM) — can be seen as a variety of IDM.

We have illustrated the power of the model by means of a few example queries expressed in an relational algebra extended to handle indexical relations. The algebra was chosen because its gives a better “flavor” of how you can cut and paste tables together. In fact, we can also define a multi-sorted calculus, similar to the language L _h discussed in [120], with variables over ordinary domains as well as over each type of index. Since L _h is shown in [120] to be more powerful than any ungrouped language for historical databases, it is reasonable to base our indexical calculus on the same framework. However, as [120] also points out, since there is as yet no known historical algebra equivalent in power to L _h the issue of the completeness of an indexical algebra remains an open one as well.