A Goal and Ontology Based Approach for Generating ETL Process Specifications

• Main Author: Azman, Ta'a
• Format: Thesis
• Language: English; English
• Published: 2012
• Subjects: QA71-90 Instruments and machines
• Online Access: https://etd.uum.edu.my/3351/1/AZMAN_TA%27A.pdf; https://etd.uum.edu.my/3351/2/AZMAN_TA%27A.pdf; https://etd.uum.edu.my/3351/; http://sierra.uum.edu.my/record=b1242564~S1
• Institution: Universiti Utara Malaysia

Data warehouse (DW) systems development involves several tasks such as defining requirements, designing DW schemas, and specifying data transformation operations. Indeed, the success of DW systems is very much dependent on the proper design of the extracting, transforming, and loading (ETL) processes. However, the common design-related problems in the ETL processes such as defining user requirements and data transformation specifications are far from being resolved. These problems are due to data heterogeneity in data sources, ambiguity of user requirements, and the complexity of data transformation activities. Current approaches have limitations on the reconciliation of DW requirement semantics towards designing the ETL processes. As a result, this has prolonged the process of the ETL processes specifications generation. The semantic framework of DW systems established from this study is used to develop the requirement analysis method for designing the ETL processes (RAMEPs) from the different perspectives of organization, decision-maker, and developer by using goal and ontology approaches. The correctness of RAMEPs approach was validated by using modified and newly developed compliant tools. The RAMEPs was evaluated in three real case studies, i.e., Student Affairs System, Gas Utility System, and Graduate Entrepreneur System. These case studies were used to illustrate how the RAMEPs approach can be implemented for designing and generating the ETL processes specifications. Moreover, the RAMEPs approach was reviewed by the DW experts for assessing the strengths and weaknesses of this method, and the new approach is accepted. The RAMEPs method proves that the ETL processes specifications can be derived from the early phases of DW systems development by using the goal-ontology approach.