Although collaborative research is vital to innovation, identifying suitable partners is challenging due to fragmented data and limited matching methods. Traditional approaches often focus narrowly on patent-based similarity, which excludes non-patenting institutions and fails to consider the specific intent of research projects. We present a novel hybrid system for matching research partners that combines semantic retrieval with structured technology classification. Our approach integrates three main components: (1) a semantic search framework that matches project descriptions to heterogeneous institutional data sources; (2) a two-step IPC-based profiling system that uses semantic retrieval and LLM refinement; and (3) a scoring mechanism that balances similarity, complementarity, and semantic proximity. This strategy enables context-aware, cross-sector partner recommendations tailored to the specific intent of research initiatives. A software prototype serves as an evaluation framework reviewed by experts. Preliminary evaluation shows a success rate of 69.2% at 10, indicating the practical utility of our hybrid approach.     «

Although collaborative research is vital to innovation, identifying suitable partners is challenging due to fragmented data and limited matching methods. Traditional approaches often focus narrowly on patent-based similarity, which excludes non-patenting institutions and fails to consider the specific intent of research projects. We present a novel hybrid system for matching research partners that combines semantic retrieval with structured technology classification. Our approach integrates thre...     »