Thesis author: Siegfried Pfaff
Legacy systems continue to support business critical processes while becoming increasingly difficult to maintain, integrate, and evolve. Research on their modernization provides a wide range of methods, techniques, technical issues, quality characteristics, process activities, risks, and technical debt concepts, however this knowledge is distributed across different publications and ontology fragments. This thesis investigates how a structured ontology engineering approach can be applied to model the technical aspects of theoretical legacy system modernization and develops a lightweight domain ontology that integrates these concepts into a single traceable conceptual model. The ontology was derived through a literature based and iterative design process consisting of scope definition, concept extraction and consolidation, reuse of existing ontology fragments, taxonomic organization, semantic relationship modeling, competency question based requirements, and evaluation. Existing software quality and technical debt ontologies as well as modernization classifications were reused where appropriate, while modernization specific concepts and semantic relations were added to connect the different areas. 21 competency questions were formulated as ontology requirements and assessed in a practitioner oriented survey with eight participants. The final ontology was implemented in yEd as a GraphML model containing 291 nodes and 320 directed edges, including 125 thesis specific integration and extension concepts. The evaluation combines competency question mapping, structural and traceability review, and two published industrial modernization cases. The results show broad conceptual coverage of the defined requirements. Some questions are represented through direct or short paths, whereas others require multistep traversal across ontology branches. The case studies largely matched the existing vocabulary and led to focused refinements consisting of two additional concepts and seven labeled relations. The thesis contributes a reusable conceptual reference model that connects legacy systems, technical issues, modernization methods and techniques, process activities, software quality, technical debt, and modernization risks. Its main limitations are the lightweight GraphML representation, the small practitioner sample, and the limited number of case studies. Future work includes formal OWL implementation, executable SPARQL queries, broader expert validation, and further industrial case based refinement.
