Fragmentation of Object Oriented and Semi-Structured Data

Abstract

Due to the increasing demand to provide data from databases for servicing web-based applications and the de facto distribution of data on the web the problem arises to generalize database distribution techniques from relational to object oriented and semi-structured data. The major difference between these two is that object orientation refers to a predefined schema, whereas semi-structured data is meant to be schema-less. The paper generalizes horizontal and vertical fragmentation first to an object oriented datamodel, then to semi-structured data and XML. In this context, the paper discusses the problem of preservation of unique object identifiability with respect to fragmentation. Furthermore, splitting will be introduced as a a third kind of fragmentation.