Dynamic Pruning for Recursive Joins

What is it about?

We have developed "dynamic pruning technology" that significantly improves the search speed of graph structure data (hereinafter referred to as "graph structure data"), which has complex interconnections, in order to accelerate big data analysis. Conventionally, when performing data analysis within a database, the process of sequentially tracing graph structure data was performed using a procedure called recursive query processing, which required repeatedly reading unnecessary data, resulting in a decrease in search speed. Our technology reduces unnecessary data retrieval by accurately identifying the range of data to be read next in real time based on information obtained during the recursive query process, thereby significantly improving search speed.

Featured Image

Photo by Simon Kadula on Unsplash

Photo by Simon Kadula on Unsplash

Why is it important?

As the use of AI and big data advances rapidly, the importance of data processing technology is increasing, and there is a growing need for more efficient data search, which is directly linked to improving AI performance and solving social issues. In databases that store vast amounts of data, it is essential to efficiently represent the complex relationships between data points. Graph-structured data, which enables this, is widely used in various analytical tasks such as traffic route search, e-commerce product recommendations, product quality control, medical data analysis, and fraud detection. In a verification study targeting product shipment decisions in the manufacturing industry, we confirmed that data search speed can be improved by up to 135 times compared to conventional methods. This is expected to contribute to improving the quality of traceability by accelerating graph structure data analysis tasks such as tracking components throughout the entire process from product design to manufacturing, distribution, and maintenance.

Perspectives