An Extended Conceptual Modeling for ETL Processes in Privacy Preserving Data Mining
Prathibha Kiran, Sathish Kumar S., N. P. Kavya · International Journal of Future Computer and Communication · 2012
Extraction Transformation Load (ETL) is an important component of data warehouse. Extraction deals with the retrieval of data from various sources. These sources of data can be of different format and different location in space (1). Each of these locations must be analyzed before it is loaded on to the data warehouse. The most important and complex part of ETL is the Transformation phase. Transformation can be visualized as the change that must be made on to the source data before it is loaded on to the data warehouse. There are various methods that are used for transformation. Usual transformation include deletion of a column, conversion of values from one representation to the other, aggregation of values for better retrieval, removal of null values, converting multiple representations in to a single format and joining of multiple tables. The last phase is the loading phase which copies the information from the transformed data to the warehouse. The overall process consists of three major steps first identification of the source targets second is the actual data representation also called conversion and the last phase is moving resultant to the target location. These tasks can be done in intermediate stages or as a single logical execution. Each of these stages indicated separately will help improve the overall representation of Data Warehouse. The initial design of conceptual schema plays an important role and also propagates to other stages. The complexity of extraction transformation and load lies in the mapping of data from source to destination. As indicated by Alkis Simitsis (1) the environment of ETL process can be shown in fig 1. The left side we have source representation where in information is