BACKGROUND

The problem of handling "slowly changing dimensions" was mentioned by Kimball (1996), who suggested some partial solutions, like timestamping dimension tuples with their validity intervals. This proposal, however, does not consider schema versioning.

A work carried out by Bliujute et al. (1998) at the Time Center at the University of Arizona analyzes the performance of several SQL queries over three different approaches to the star schema: (a) "time series" fact tables; (b) "event" fact tables; (c) dimensions timestamped in the way proposed by Kimball (1996). This work was, to our knowledge, the first approach to the problem of temporal OLAP. Our work goes further, as we propose a model and a query language to address temporal issues at a higher abstraction level.

More recently, a multidimensional model for handling complex data was introduced by Pedersen & Jensen (1999) where the temporal aspect is considered a modeling issue, and is addressed in conjunction with other data modeling problems. At the time this is being written, we are not aware of any other work proposing a data warehouse evolution framework or a temporal query language for OLAP like the ones addressed in this chapter.

Recent work on maintenance of temporal views by Yang & Widom (2000) presents a view definition language operating over non-temporal data sources, along with techniques for maintaining temporal views. This work focuses on the data sources and on how a set of temporal views are obtained and maintained. We will not deal with these issues in this chapter.

Mendelzon & Vaisman (2000) discussed the need for a temporal query language for OLAP, introducing TOLAP, along with a temporal multidimensional data model.