Regresar

Fragmenting Data Strategies to Scale Up the Knowledge Graph Creation

Abstract:

In recent years, the exponential growth of data has necessitated a unified schema to harmonize diverse data sources. This is where knowledge graphs (KGs) come into play. However, the creation of KGs introduces new challenges, such as handling large and heterogeneous input data and complex mappings. These challenges can lead to reduced scalability due to the significant memory consumption and extended execution times involved. We present KatanaG, a framework designed to streamline KG creation in complex scenarios, including large data sources and intricate mapping. KatanaG optimizes memory usage and execution time. When applied alongside various KG creation engines, our results indicate that KatanaG can improve the performance of these engines, by reducing execution time by up to 80% and achieve 70% memory savings.