An evaluation of a metamodel for documenting data transformations in the ALPHA Network

The network for Analysing Longitudinal Population based HIV/AIDS data on Africa (ALPHA) brings together ten African collaborating research institutions conducting community-based HIV surveillance. At the heart of the network's activities is extensive after-the-fact data harmonisation processes comprising transformations of data from disparate sources into predefined uniform formats. While the documentation of the source data is generally understood, the documentation of data transformations involved in creating the ALPHA harmonised datasets in still shrouded in mystery. In an effort to address this issue, our study evaluates the applicability of a generic and customisable framework for describing extract, transform, and Load (ETL) scenarios for documenting ALPHA transformations. We seek to examine whether the framework covers the common ALPHA data harmonisation scenarios. Further, we assess its capabilities regarding facilitation of reproduction of ALPHA data transformations processes. We will also consider the potential for further formalisation of the ALPHA adapted framework within metadata standards such as the Data Documentation Initiative (DDI 4) model.