Michael Scriney - Transfer Talk - 20th August 2015

Video Category: 
Transfer Talk
Michael Scriney

Title: Using the NoSQL Model to support DWARF Cubes in XML Data Mining

Supervisor: Dr. Mark Roantree

Abstract:

Modern data analysis depends heavily on Data Warehouses and Data Cubes. Data Warehouses provide the means of storing large amounts of data and Cubes facilitate both multidimensional analytical processing and the construction of data mining models. However, Cubes take a large amount of time to construct and have a very large storage overhead. While there have been many optimisation techniques for relational warehouses and cubes, there is less so for non-relational sources such as XML and indeed, the construction and refreshing of XML Cubes remains an ongoing problem. One promising approach to the management of XML Cubes employs a construct known as DWARF but there has been no work on a suitable storage model for DWARF Cubes or Schemas, nor for an update mechanism to refresh the Cube to keep data mining models current. In this research, we propose to use the DWARF structure in conjunction with a NoSQL model to provide for XML-DWARF construction and the efficient maintenance of DWARF Cubes.