Abstract
Data warehouses (DWs) are used for storing and analyzing high volumes of historical data. The structure of DWs is usually represented as a star schema consisting of fact and dimension tables. A fact table contains numeric data called measures (e.g., quantity). Dimensions are used for exploring measures from different analysis perspectives (e.g., according to products). They usually contain hierarchies required for online analysis processing (OLAP) systems in order to dynamically manipulate DW data. While traversing hierarchy, two operations can be executed: the roll-up operation, which transforms detailed measures into aggregated data (e.g., daily into monthly sales); and the drill-down operation, which does the opposite.
TopBackground
To the best of our knowledge, very few proposals address the issue of conceptual modeling for SDWs (Ahmed & Miquel, 2005; Bimonte, Tchounikine & Miquel, 2005; Jensen, Klygis, Pedersen & Timko, 2004; Pestana, Mira da Silva & Bédard, 2005). Some of these models include the concepts presented in Malinowski and Zimányi (2004), as explained in the next section; other models extend nonspatial multidimensional models with different aspects such as imprecision in location-based data (Jensen et al., 2004) or continuous phenomena (e.g., temperature or elevation) (Ahmed & Miquel, 2005).
Other models for SDWs use the logical relational representation based on the star/snowflake schemas. These proposals introduce concepts of spatial dimensions and spatial measures (Fidalgo, Times, Silva & Souza, 2004; Rivest, Bédard & Marchand, 2001; Stefanovic, Han & Koperski, 2000); however, they impose some restrictions on the model, as discussed in the next section.
We consider that a conceptual multidimensional model with spatial support should not only include dimensions, hierarchies, and measures, but should also refer to various aspects that are not present in conventional multidimensional models related to particularities of spatial objects.
Spatial objects correspond to real-world entities for which the application needs to keep their spatial characteristics. Spatial objects consist of a thematic (or descriptive) component and a spatial component. The thematic component describes general characteristics of spatial objects (e.g., name) and is represented using traditional DBMS data types (e.g., integer, string, date). The spatial component includes its geometry that can be of type point, line, surface, or a collection of them.
Key Terms in this Chapter
Spatial Measure: An attribute of a (spatial) fact relationship that can be represented by a geometry or calculated using spatial operators.
Spatial Hierarchy: One or several related levels where at least one of them is spatial.
Spatial Level: A type defining a set of attributes, keeping track of the spatial extent of its instances (members).
Spatial Data Warehouse: A data warehouse that includes spatial dimensions, spatial measures, or both, thus allowing spatial analysis.
Multidimensional Model: A model for representing the information requirements for data warehouse and OLAP applications. It includes facts, measures, dimensions, and hierarchies.
Spatial Fact Relationship: An n-ary relationship between two or more spatial levels belonging to different spatial dimensions.
Spatial Dimension: An abstract concept for grouping data that share common semantics within the domain being modeled. It contains a spatial level or one or more spatial hierarchies.