Domain Modeling Using Qualitative Data Analysis

Kaufmann A (2021)


Publication Language: English

Publication Type: Thesis

Publication year: 2021

URI: https://opus4.kobv.de/opus4-fau/files/16736/AndreasKaufmannDissertation.pdf

Abstract

The creation of domain models from qualitative input relies heavily on experience. An uncodified ad-hoc modeling process is still common and leads to poor documentation of the requirements analysis.

In this thesis, we present a novel method for domain analysis based on qualitative data analysis (QDA). The method helps identifying inconsistencies, ensures a high degree of completeness, and inherently provides traceability from analysis results back to stakeholder input.

In our approach, the QDAcity-RE method, the research process of theory building facilitates domain analysis within the requirements elicitation phase of a software development project.

We show how an iterative process of concurrent data collection and analysis can be applied to requirements engineering (RE), including open, axial, and selective coding of qualitative data.

The traceability of domain model elements back to original statements by stakeholders, generated by our method, does not have to be created and maintained separately after the fact. The traces are documented in an analysis artifact called the code system, which evolves iteratively with the analysis process. The code system can act as a universal model from which a variety of artifacts can be derived, describing both behavioral and structural aspects. This thesis focuses on the creation of conceptual domain models using QDAcity-RE, but peripherally also demonstrates this capability through the generation of behavioral models and a software requirements specification.

We applied and evaluated our method for domain modeling in four exploratory projects in the domains of medical imaging diagnostics, railway systems, HR development, and qualitative research.

We show that by applying QDA to domain analysis, structural elements and relationships needed to derive a UML class diagram can be extracted from a code system based on interviews with domain experts. Constant comparison and theoretical sampling assist in integrating differing domain descriptions into an abstract model.

While the analysis process still requires interpretations and modeling decisions, our method provides more guidance than existing domain analysis approaches and a thorough documentation of these decisions. In addition, codes and memos ensure traceability between the original data and the derived model and assist in connecting several RE artifacts, ensuring a high degree of inter-model consistency.

We validated our claim that QDAcity-RE helps an analyst gain a deeper understanding of a problem domain through a controlled experiment.

Authors with CRIS profile

How to cite

APA:

Kaufmann, A. (2021). Domain Modeling Using Qualitative Data Analysis (Dissertation).

MLA:

Kaufmann, Andreas. Domain Modeling Using Qualitative Data Analysis. Dissertation, 2021.

BibTeX: Download