Design and representation of complex objects in database systems

Abstract

In recent decades, applications like geospatial, genomic, and multimedia have made use of very large and diverse application objects such as spatial networks and protein structures. These objects are complex in the sense that they are highly structured and of variable size. Storing, accessing and manipulating them in a standard and efficient manner is very challenging. The state-of-the-art solutions handle such objects by using file system formats like HDF and XML, serialization technique like Protocol Buffers and BLOB data type in databases. However, specialized file format solutions lack any well established database system features, and neither a uniform concept nor mechanisms exist for supporting complex objects for BLOBs. In this article, a novel and database-friendly framework of specifying and interpreting complex objects is proposed. Empirical studies have shown that our approach outperforms prevailing methods with efficient processing time and less storage consumption.