Design science methodology is an iterative and problem-solving approach used in research to develop innovative solutions for practical problems. It is commonly applied in areas such as information systems, engineering, and computer science. The primary goal of design science methodology is to create artifacts, such as models, frameworks, or prototypes, that address specific real-world problems and contribute to knowledge in a particular domain.
The methodology involves a cyclical process of problem identification, problem analysis, artifact design and development, and evaluation. It emphasizes the importance of rigorous research methods combined with practical problem-solving techniques. Design science methodology is driven by the idea of creating useful and effective solutions that can be applied in practice, rather than solely focusing on theorizing or studying existing phenomena.
In this approach, researchers actively engage with stakeholders, gather requirements, and design artifacts that can be implemented and tested. The evaluation phase is crucial, as it assesses the effectiveness, efficiency, and practicality of the developed artifact, allowing for further refinement or iteration. The ultimate goal is to contribute to knowledge by providing practical solutions and insights that can be shared with the academic and professional communities.
Design science methodology offers a systematic and structured framework for problem-solving and innovation, combining theoretical knowledge with practical application. By following this methodology, researchers can generate actionable solutions that address real-world problems and have a tangible impact on practice.
The two major components that represent a design science activity for any research project are two mandatory requirements:
- The object of the research is an artifact in this context.
- The research comprises two main actions: designing and investigating the artifact within the context. To achieve this, a thorough examination of the literature was conducted to create a process model. The process model consists of six activities that are sequentially organized. These activities are further described and visually presented in Figure 11.
Figure 1: DSRM Process Model [1]
Problem Identification and Motivation
The initial step of problem identification and motivation involves defining the specific research problem and providing justification for finding a solution. To effectively address the problem’s complexity, it is beneficial to break it down conceptually. Justifying the value of a solution serves two purposes: it motivates both the researcher and the research audience to pursue the solution and accept the outcomes, and it provides insight into the researcher’s understanding of the problem. This stage necessitates a solid understanding of the current state of the problem and the significance of finding a solution.
Solution Design
Determining the objectives of a solution is a crucial step in the solution design methodology. These objectives are derived from the problem definition itself. They can be either quantitative, focusing on improving existing solutions, or qualitative, addressing previously unexplored problems with the aid of a new artifact [44]. The inference of objectives should be rational and logical, based on a thorough understanding of the current state of problems, available solutions, and their effectiveness, if any. This process requires knowledge and awareness of the problem domain and the existing solutions within it.
Design Validation
In the process of design validation, the focus is on creating the actual solution artifact. This artifact can take various forms such as constructs, models, methods, or instantiations, each defined in a broad sense [44]. This activity involves identifying the desired functionality and architecture of the artifact, and then proceeding to develop the artifact itself. To successfully transition from objectives to design and development, it is essential to have a strong understanding of relevant theories that can be applied as a solution. This knowledge serves as a valuable resource in the design and implementation of the artifact.
Solution Implementation
In the implementation methodology, the main objective is to showcase the effectiveness of the solution artifact in addressing the identified problem. This can be achieved through various means such as conducting experiments, simulations, case studies, proofs, or any other suitable activities. Successful demonstration of the artifact’s efficacy requires a deep understanding of how to effectively utilize the artifact to solve the problem at hand. This necessitates the availability of resources and expertise in employing the artifact to its fullest potential for solving the problem.
Evaluation
The evaluation methodology in the context of anomaly detection focuses on assessing how well the artifact supports the solution to the problem. This involves comparing the intended objectives of the anomaly detection solution with the actual results observed during the artifact’s demonstration. It requires understanding relevant evaluation metrics and techniques, such as benchmarking the artifact’s performance against established datasets commonly used in the anomaly detection field. At the end of the evaluation, researchers can make informed decisions about further improving the artifact’s effectiveness or proceeding with communication and dissemination of the findings.
[1] Noseong Park, Theodore Johnson, Hyunjung Park, Yanfang (Fanny) Ye, David Held, and Shivnath Babu, “Fractyl: A platform for scalable federated learning on structured tables,” Proceedings of the VLDB Endowment, vol. 11, no. 10, pp. 1071–1084, 2018.