Generating Software Architecture Description from Source Code using Reverse Engineering and Large Language Model

Overview

The paper presents a method to generate Software Architecture Descriptions (SADs) from source code using reverse engineering and large language models (LLMs). This approach aims to address the issues of missing or outdated SADs by providing a scalable and maintainable alternative.

Key Insights

• Semi-automated SAD Generation: The method integrates reverse engineering with LLMs to generate SADs, reducing reliance on manual documentation.
• Component and Behavior Representation: It recovers static and behavioral views, including component diagrams and state machine diagrams.
• Reduced Manual Effort: The approach significantly reduces the manual effort required to maintain architectural documentation.

BFSI Relevance

• Why Relevant: Accurate and up-to-date software architecture is crucial for maintaining complex financial systems.
• Primary Sector: Financial Services
• Subsectors: Software Development, IT Infrastructure
• Actionable Implications: BFSI professionals should consider adopting such automated tools to enhance system documentation and reduce maintenance costs.