LLM-Driven Automatic Knowledge Transfer for Promoting Organizational Software Upcycling

Software upcycling is a methodology aimed at improving the efficiency of the software development process by reusing existing software resources. Previous research introduced the SUCCEED system, which allows development organizations to share upcycling achievements manually as cases. This study proposes an automated knowledge transfer method that leverages large language models (LLMs) for data transformation and filtering, enabling the integration of external community development insights into internal organizational practices as upcycling cases. The proposed method is implemented and evaluated using Stack Overflow. The results demonstrate that LLMs enable accurate transformation of case data, and the proposed filtering approach effectively integrates external knowledge into internal organizational contexts.