1

Model-Driven Metadata for OLAP Cubes fromthe Conceptual Modeling of Data Warehouses

Author: Jesus PardilloJose-Norberto Mazon

Juan TrujilloPresenter: Michael Shaobo Wang

2

Outline

Current overlooking Research challenge Automatic generation of OLAP

metadata Implementation Conclusion & future plan

3

What does multidimensional modeling do

The development of a data warehouse is based on the multidimensional modeling Intuitive Semantically rich Repository structures Analysis structures

4

Multidimensional Modeling

5

What should be done

Theoretically, once a conceptual multidimensional model is defined, two kinds of logical models should be derived

A model of the data repository which determines the required database metadata to structure the data warehouse

A model of the data cubes which contains the necessary metadata to allow end-user tools to query the data warehouse in a suitable format

6

Two kinds of metadata should be derived

7

Current situation from industry

The current commercial data warehouse development platforms derive the data-cube metadata from logical models in a manually or semi automatically way vendor-specific require certain tedious post processing (e.g., renaming data

entities or defining aggregation hierarchies) proceeding is prone-to-fail requires great amount of effort

The global cost of the data-warehouse development is increased.

8

Current situation

9

Current overlooking

Unfortunately and surprisingly, current approaches based on multidimensional modeling only focus on deriving the database metadata from a conceptual multidimensional model, thus overlooking the derivation of the necessary data-cube metadata.

10

Multidimensional Modeling

11

Outline

Current overlookingResearch challengeAutomatic generation of OLAP

metadataImplementationConclusion & future plan

12

Research challenge

The generation of data-cube metadata should be derived together with database metadata in an integrated way without any reference to a specific software platform or technology.

13

Outline

Current overlooking Research challenge Automatic generation of OLAP metadata Implementation Conclusion & future plan

14

Approach overview

The author’s approach employs a three-layer model-transformation architecture based on conceptual, logical, and physical design phases

Firstly, they specify the information requirements into a conceptual multidimensional model

From this model, the logical model of the data-warehouse repository and OLAP data cubes are derived, both represented in a vendor-independent format

Finally, the physical implementation for a specific vendor platform is obtained

15

The model mappings will be discussed

Specifically, the author discussed the necessary model mappings between the conceptual modeling framework and the logical OLAP model represented in CWM

16

Common Warehouse Metamodel (CWM)

The specification of logical OLAP metadata is carried out by means of the “common warehouse metamodel”(CWM), an industry standard for information management, specifically oriented to the metadata interchange.

17

Running example: automobile sales

The conceptual multidimensional model of automobile sales

18

Model mappings

From the conceptual modeling framework to the logical OLAP model, four kinds of mapping was described:

Mapping facts and measures into OLAP cubesMapping dimensions and aggregation levels.Mapping aggregation hierarchies.Mapping dimension attributes into level-based attributes.

19

OLAP metadata mapped in CWM to specify the customer dimension

20

Mapping the Name attribute of Region and State levels of Customers

21

Mapping result

In short, with the designed mappings, we can obtain the OLAP metadata to query the database structures with an integrated and vendor-neutral approach

22

Outline

Current overlooking Research challenge Automatic generation of OLAP metadata Implementation Conclusion & future plan

23

Language and technologiesPIM

PSM PSM

UML

CWMCWM

MDA

QVT QVT

24

Implementation in Eclipse

Several developments to support the MDA standards

The Model Development Tools (MDT) for modeling with UML

MediniQVT for specifying declarative QVT relationsSmartQVT for imperative QVT relations

25

IDE based on Eclipse of the case study

UML QVT CWM

26

Outline

Current overlooking Research challenge Automatic generation of OLAP metadata Implementation Conclusion & future plan

27

Conclusion

In this paper, the author discuss the involved conceptual–logical mappings and present a set of QVT relations to formalize the mapping between the conceptual modeling framework for data warehouses and the analysis structures of the OLAP metadata represented in CWM, a vendor-neutral standard for metadata interchange.

28

Future plan

Extend this approach to consider degenerated facts and dimensions, also investigate how their logical OLAP counterparts are implemented.

The derivation of specific OLAP metadata for non-traditional OLAP tools such as in mobile environments, as well as their enrichment with advanced customization issues such as OLAP preferences.

Apart from OLAP metadata, other analysis techniques, such as data mining, should be also considered.

In addition, we will further investigate the existing mappings between data-cube and database structures.

29

References 1. Jarke, M., Lenzerini, M., Vassiliou, Y., Vassiliadis, P.: Fundamentals of DataWarehouses. Springer, Heidelberg (2000) 2. Inmon, W.H.: Building the Data Warehouse. Wiley, Chichester (2005) 3. Kimball, R., Ross, M.: The Data Warehouse Toolkit: The Complete Guide to Dimensional Modeling. Wiley, Chichester (2002) 4. H¨usemann, B., Lechtenb¨orger, J., Vossen, G.: Conceptual data warehouse modeling. In: DMDW, p. 6 (2000) 5. Rizzi, S., Abell′o, A., Lechtenb¨orger, J., Trujillo, J.: Research in data warehouse modeling and design: dead or alive?. In: DOLAP, pp.

3–10 (2006) 6. Golfarelli, M., Maio, D., Rizzi, S.: The Dimensional Fact Model: A Conceptual Model for Data Warehouses. Int. J. Cooperative Inf. Syst.

7(2-3), 215–247 (1998) 7. Abell′o, A., Samos, J., Saltor, F.: YAM2: a multidimensional conceptual model extending UML. Inf. Syst. 31(6), 541–567 (2006) 8. Prat, N., Akoka, J., Comyn-Wattiau, I.: A UML-based data warehouse design method. Decis. Support Syst. 42(3), 1449–1473 (2006) 9. Pardillo, J., Maz′on, J.N., Trujillo, J.: Towards the Automatic Generation of Analytical End-user Tool Metadata for Data Warehouses. In:

BNCOD, pp. 203–206 (2008) 10. B′ezivin, J.: Model Driven Engineering: An Emerging Technical Space. In: GTTSE, pp. 36–64 (2006) 11. Chaudhuri, S., Dayal, U.: An Overview of Data Warehousing and OLAP Technology. SIGMOD Record 26(1), 65–74 (1997) 12. Maz′on, J.N., Trujillo, J.: An MDA approach for the development of data warehouses. Decision Support Systems 45(1), 41–58 (2008) 13. Hahn, K., Sapia, C., Blaschka, M.: Automatically Generating OLAP Schemata from Conceptual Graphical Models. In: DOLAP, pp. 9–

16 (2000) 14. Object Management Group (OMG): Catalog of OMG Specifications (March 2008), http://www.omg.org/technology/documents/spec

catalog.htm 15. Luj′an-Mora, S., Trujillo, J., Song, I.Y.: A UML profile for multidimensional modeling in data warehouses. Data Knowl. Eng. 59(3), 725–

769 (2006) 16. Giovinazzo, W.A.: Object-Oriented Data Warehouse Design: Building A Star Schema. Prentice Hall, Englewood Cliffs (2000) 17. Cuzzocrea, A., Furfaro, F., Sacc`a, D.: Hand-OLAP: A System for Delivering OLAP Services on Handheld Devices. In: ISADS, pp.

80–87 (2003) 18. Maniatis, A.S.: The Case for Mobile OLAP. In: Lindner, W., Mesiti, M., T¨urker, C., Tzitzikas, Y., Vakali, A.I. (eds.) EDBT 2004. LNCS,

vol. 3268, pp. 405–414. Springer, Heidelberg (2004) 19. Rizzi, S.: OLAP preferences: a research agenda. In: DOLAP, pp. 99–100 (2007) 20. Zubcoff, J.J., Pardillo, J., Trujillo, J.: Integrating Clustering Data Mining into the Multidimensional Modeling of Data Warehouses with

UML Profiles. In: Song, I.- Y., Eder, J., Nguyen, T.M. (eds.) DaWaK 2007. LNCS, vol. 4654, pp. 199–208. Springer, Heidelberg (2007)

30

Q & A