Conference Paper

Handling Evolutions in Multidimensional Structures.

DOI: 10.1109/ICDE.2003.1260823 Conference: IEEE 19th International Conference on Data Engineering (ICDE), At Bangalore, India
Source: DBLP


Building multidimensional systems requires gathering data from heterogeneous sources throughout time. Then, data is integrated in multidimensional structures organized around several axes of analysis, or dimensions. But these analysis structures are likely to vary over time and the existing multidimensional models do not (or only partially) take these evolutions into account. Hence, a dilemma appears for the designer of data warehouses: either keeping trace of evolutions therefore limiting the capability of comparison for analysts, or mapping all data in a given version of the structure that entails alteration (or even loss) of data. We propose a novel approach that offers another alternative, allowing to track history but also to compare data, mapped into static structures. We define a conceptual model and provide possible logical adaptations to implement it on current commercial OLAP systems. At last, we present the global architecture that we used for our prototype.

Download full-text


Available from: Yvan Bédard
  • Source
    • "However, they limit the usefulness of these temporality types for only active data warehouses, i.e., for data warehouses that include event-conditionaction rules (or triggers). The inclusion of temporal support raise many issues, such as efficient temporal aggregation of multidimensional data (Moon, Vega, & Immanuel, 2003), correct aggregation in presence of data and schema changes (Body et al., 2003; Eder, Koncilia, & Morzy, 2002; Wrembel & Bebel, 2007; Mendelzon & Vaisman, 2003; Golfarelli, Lechtenbörger, Rizzi, & Vossen, 2006), or temporal view materialization from non-temporal sources (Yang & Widom, 1998). Even though the works related to schema and data changes define models for temporal data warehouses, what is still missing is a conceptual model that allows decision-making "
    [Show abstract] [Hide abstract]
    ABSTRACT: Current data warehouses include a time dimension that allows one to keep track of the evolution of measures under analysis. Nevertheless, this dimension cannot be used for indicating changes to dimension data. Based on the research in temporal databases, in this chapter we present a conceptual model for designing temporal data warehouses. The model supports different temporality types, i.e., lifespan, valid time, transaction time coming form source systems, and loading time, generated in a data warehouse. This support is used for representing time-varying levels, dimensions, hierarchies, and measures.
    Full-text · Chapter · Jan 2009
  • Source
    • "This definition is very similar to the one given in [13], the difference is that the first one supports versioning. In [14] a method to support data and structure versions of dimensions is proposed. The method allows tracking history and comparing data using temporal modes of presentation that is data mapping into the particular structure version. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Schemata of data warehouses often need to be adapted because of evolving business requirements or changes in data sources. To accumulate the history of schemata and data, it is possible to maintain multiple versions of data warehouse schemata. We propose the metadata models to store the data about data warehouse logical and physical schemata and their versions and specification of reports on multiple versions of data warehouse schemata. We also present the approach to build SQL queries using the proposed metadata. A user can define a report using elements of a data warehouse schema, and queries can be generated on one or several versions of schemata according to the proposed approach.
    Full-text · Conference Paper · Jun 2008
  • Source
    • "First, having valid time and lifespan support, users can analyze measures taking into account changes in dimension data. Second, these temporality types help implementers to develop procedures for correct measure aggregation during roll-up operations in the presence of changes in dimension data [9] [17] [40] [62]. Finally, transaction time is important for traceability applications, for example, for fraud detection, when the changes to data in operational databases and the time when they occurred are required for investigation processes. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Current data warehouse and OLAP models include a time dimension that, like other dimensions, is used for grouping purposes (using the roll-up operation) or in a predicate role (using the slice-and-dice operation). The time dimension also indicates the time frame for measures (for example, in order to know how many units of a product were sold in March 2007). However, the time dimension cannot be used to keep track of changes in other dimensions, for example, when a product changes its ingredients or its packaging. Consequently, the “nonvolatile" and “time-varying" features included in the definition of a data warehouse (Sect. 2.5) apply only to measures, and this situation leaves to applications the responsibility of representing changes in dimensions. Kimball et al. [147] proposed several solutions for this problem in the context of relational databases, the slowly changing dimensions. Nevertheless, these solutions are not satisfactory, since they either do not preserve the entire history of the data or are difficult to implement. Further, they do not take account of all research that has been done in the field of temporal databases.
    Full-text · Article · Jan 2008
Show more