GraphAI CEO and KAIST School of Computing Professor Minsoo Kim’s research team has developed ‘Chimera,’ a new database system that fully integrates relational DB and graph DB to enable significantly more efficient execution of graph-relational queries. Chimera addresses the performance degradation and memory shortage issues of existing systems by introducing a newly designed ‘dual-store architecture’ and ‘traverse-join operator’ across both the storage layer and the query processing layer.
As a result, Chimera recorded query processing performance that is at least 4x and up to 280x faster than existing systems on international performance benchmark standards, demonstrating world-class capabilities. This research is expected to find wide application across various industries — including AI agents, finance, and e-commerce — by enabling the efficient execution of graph DB’s strengths in analyzing complex interconnected information alongside traditional relational DB functions within a single framework.
GraphAI engineer 박정호 participated as the second author, contributing to the technical completeness of the research. The results were presented at ‘VLDB,’ one of the world’s most prestigious international academic conferences in the database field.
Read more : https://view.asiae.co.kr/article/2025090808190435902