25 Jan 2024 | Tobias Eisenreich, Sandro Speth, Stefan Wagner
The paper "From Requirements to Architecture: An AI-Based Journey to Semi-Automatically Generate Software Architectures" by Tobias Eisenreich addresses the challenge of designing domain models and software architectures, which are crucial for ensuring the quality of software systems. The authors propose a semi-automated method to generate software architecture candidates based on requirements using artificial intelligence (AI) techniques. This approach aims to reduce the time and effort required for manual architecture design, which is often limited by the architect's domain knowledge and experience.
The paper outlines a six-step process: generating a domain model and use-case scenarios from textual requirements, refining these models and scenarios, deriving multiple software architecture candidates, evaluating and comparing the candidates, refining the candidates, and selecting the best-fitting candidate. The use of large language models (LLMs) is highlighted as a key component, capable of generating initial domain models and use-case scenarios with reasonable accuracy. The authors also plan to evaluate the quality of the generated architecture models and the efficiency and effectiveness of their proposed process through qualitative studies.
The related work section reviews existing methods for architecture derivation and evaluation, noting that while some approaches are semi-automatic, they still require significant manual effort. The exploratory analysis section demonstrates the potential of LLMs in generating domain models, though further adjustments are needed. The planned evaluation section details two studies: one using reference architectures and another conducting an industrial field study to assess the practicality and effectiveness of the proposed process.
In conclusion, the paper presents a comprehensive vision for a semi-automated architecture generation process that can improve the quality and efficiency of software architecture design.The paper "From Requirements to Architecture: An AI-Based Journey to Semi-Automatically Generate Software Architectures" by Tobias Eisenreich addresses the challenge of designing domain models and software architectures, which are crucial for ensuring the quality of software systems. The authors propose a semi-automated method to generate software architecture candidates based on requirements using artificial intelligence (AI) techniques. This approach aims to reduce the time and effort required for manual architecture design, which is often limited by the architect's domain knowledge and experience.
The paper outlines a six-step process: generating a domain model and use-case scenarios from textual requirements, refining these models and scenarios, deriving multiple software architecture candidates, evaluating and comparing the candidates, refining the candidates, and selecting the best-fitting candidate. The use of large language models (LLMs) is highlighted as a key component, capable of generating initial domain models and use-case scenarios with reasonable accuracy. The authors also plan to evaluate the quality of the generated architecture models and the efficiency and effectiveness of their proposed process through qualitative studies.
The related work section reviews existing methods for architecture derivation and evaluation, noting that while some approaches are semi-automatic, they still require significant manual effort. The exploratory analysis section demonstrates the potential of LLMs in generating domain models, though further adjustments are needed. The planned evaluation section details two studies: one using reference architectures and another conducting an industrial field study to assess the practicality and effectiveness of the proposed process.
In conclusion, the paper presents a comprehensive vision for a semi-automated architecture generation process that can improve the quality and efficiency of software architecture design.