Multidimensional Databases: Problems and Solutions
|
|
We have presented a logical data model, based on the extension of the geographical data model, which provides the capability of manipulating complex entities and answering queries by the cooperation with MDDBs. The cooperation of spatio-aggregate data is obtained by introducing new attributes into the geographic data structure. These are called binding attributes. The main idea behind them is to retrieve information which is stored in MDDBs and can be invoked from queries formulated in GDBs.
We have shown that our approach to the common "spatial" feature provides the interoperability of spatio-aggregate databases and guarantees the autonomy of data sources. The features of our approach that distinguish it from similar works include:
-
uniform treatment of spatio-aggregate data;
-
a comprehensive aggregation facility of either geographic or summary data;
-
a non explicit different use of a specific set of operators in each database; and
-
a specific design for interoperability.
The discussion up to this point has focused upon the space feature for realizing the interoperability. However, the role of the Time feature in such a cooperation forms an interesting issue to be considered in future work.
An important assumption made in this chapter concerns the summarizability of a data cube, which means no cell contains a "not available" value, and there is no missing value. This restriction can be of practical importance in some cases of data warehousing. In many other cases the partial availability of data is realistic and consequently the comprehensive aggregation of summary data is compromised. The answering of joint aggregate queries formulated on cubes with missing and/or non-available data values in the context of the interoperability of heterogeneous data sources can create another open problem.
We discussed the problem of answering spatio-summary queries in the context of GDBs. Research that considers this question in the special light of MDDBs could raise another interesting issue to be examined.
|
|