SEOUL, September 08 (AJP) - The Korea Advanced Institute of Science & Technology (KAIST) has developed a new database system that brings together two types of databases, relational and graph, into a single, more powerful platform, the prestigious institute said Monday. The technology, called "Chimera," is expected to help create AI agents that can think more like humans by understanding not just simple facts but also complex connections between them.
Relational databases are the traditional way companies have stored data, like rows and columns in a spreadsheet. Graph databases work differently: they store data as nodes and links, making them especially good at showing relationships, such as who is friends with whom, or how a buyer, product, and seller are connected. Until now, companies often had to manage these two systems separately, which created high costs, delays, and errors when combining results.
Chimera solves this by completely integrating the two systems. It can process both relational and graph queries at the same time, without the usual slowdown or memory issues. This means AI programs can answer more complicated questions, such as "Who are the friends of this person's friends, and where do they work?" much faster and more accurately.
In international benchmark tests, Chimera was at least four times faster, and in some cases up to 280 times faster, than existing systems. It also avoids the memory shortages and outdated data problems that earlier solutions often faced.
"Data is getting more connected and complex, so the need for technology that combines graph and relational databases is growing," said Kim Min-soo, professor at KAIST's School of Computing. "Chimera addresses this need directly, and we expect it to be widely used in AI, finance, and e-commerce."
The system was presented on September 1 at the VLDB international conference, one of the world's leading events for database research. It is already being applied to "AkasicDB," a new platform being released by GraphAI, a startup founded by Professor Kim. AkasicDB is designed to power next-generation AI assistants that can search and reason in real time.
The study's first author was doctoral candidate Lee Geon-ho, with Jeong-ho Park of GraphAI as the second author. The project was supported by the Institute for Information & Communications Technology Planning & Evaluation (IITP) and the National Research Foundation of Korea.
Copyright ⓒ Aju Press All rights reserved.
View more comments