Towards a Multi-Objective Modularization Approach for Entity-Relationship Models


Legacy systems and their associated data models often evolve into large, monolithic artifacts. This threatens comprehensibility and maintainability by human beings. Breaking down a monolith into a modular structure is an established technique in software engineering. Several previous works aimed to adapt modularization also for conceptual data models. However, we currently see a research gap manifested in the absence of: (i) a flexible and extensible modularization concept for Entity Relationship (ER) models; (ii) of openly available tool support; and (iii) empirical evaluation. With this paper, we introduce a generic encoding of a modularization concept for ER models which enables the use of meta-heuristic search approaches. For the efficient application we introduce the ModulER tool. Eventually, we report on a twofold evaluation: First, we demonstrate feasibility and performance of the approach by two demonstration cases. Second, we report on an initial empirical experiment and a survey we conducted with modelers to compare automated modularizations with manually created ones and to better understand how humans approach ER modularization.

in: “ER Forum, Demo and Posters 2020”, J. Michael, V. Torres (ed.); issued by: RTWH Aachen;,, 2020, 45 - 58
Dominik Bork
Dominik Bork
Assistant Prof. Dipl.-Wirtsch.Inf.Univ. Dr.rer.pol.
Manuel Wimmer
Privatdoz. Mag.rer.soc.oec. Dr.rer.soc.oec.