Enterprise architecture analysis based on network paradigm: a framework proposal and empirical evaluation

Abstract

While the literature on Enterprise Architecture (EA) models, frameworks, and methodologies for EA implementation has many exemplars, the field is still missing mechanisms of EA analysis. EA analysis is the process which uses any technique or method to extract information from EA models about a particular concern, in order to support EA management by the experts or inform stakeholders. In this thesis, we model the EA as a complex network, a concept discussed in network science, to analyze EA structural aspects. During our exploratory study about EA network analysis (EANA), it was clear that the field was still lacking foundational aspects. First, no common language was shared by researchers. Secondly, there was no clarity about what concerns could be analyzed with network analysis initiatives and thirdly, the techniques and methods´ implementation were not clear in the papers. We solve those gaps in order to describe how to perform analysis of EA components and their relationships supported by network measures. The research approach comprehends qualitative methods such as systematic literature review, thematic analysis and design science research method. The research is conducted in three complementary and interrelated phases, aiming at first, to collect and synthesize the available knowledge about the analysis approaches existent in the literature. Next, we aim to trace a comprehensive understanding of the main concepts involved in EANA such as their analysis concerns, modeling decisions, inputs required and steps necessary to perform it. Altogether, this resulted in a set of six proposed artifacts: EANA meta-model, EANA library, EANA process, EANA data derivation strategy. Finally, we investigate the use of those artifacts, evaluating them empirically through their instantiations and/ or with the help of EA experts of three German multinational companies. The evaluation results were positive regarding, among other criteria, the efficacy and utility of the proposed artifacts in their respective contexts. As contributions, we claim the definition of the conceptual foundations of the EANA research field. Complementary, the study is not limited to the theoretical findings since it advances the understanding of empirical network analysis, whereas it offers a library of analysis initiatives, methods to derive EA data and guidelines to help experts through the analysis process (EANA process). Finally, we also add to the EANA knowledge base two new EANA methods which were also empirically evaluated. We expect that results can enhance the awareness researchers and practitioners about the EA network-based analysis´ efficacy and utility, a step necessary to develop more rationally grounded methods and tools to support the EA management considering structural aspects.