General Information

My name is Martin O'Connor. I am currently pursuing a Ph.D., in the area of dynamic labeling schemes for XML, with a particular emphasis on supporting XML updates.

I am a member of the Interoperable Systems Group (ISG) and my supervisor is Dr. Mark Roantree. The ISG research team is based in the School of Computing located in Dublin City University, Ireland.

 

 

Phd Topic:  Dynamic Labeling Schemes for XML Updates

Overview

The eXtensible Markup Language (XML) has been adopted as the new standard for data exchange on the World Wide Web. More and more information from the financial, legal, health, scientific and government domains is been processed, integrated, queried, transformed and published as XML. As the rate of XML adoption increases, there is an ever pressing need to store and maintain the information in XML format and thereby eliminating the overhead of parsing and transforming XML in and out of various data formats. At present, most modern databases provide support for storing and querying XML documents. They also support the updating of XML data at the document level, but provide limited and inefficient support for the more fine-grained (node-based) updates within XML documents. In my doctoral research, I have focused on investigating the essential problems prohibiting efficient, effective and holistic XML updates with the goal of designing solutions to overcome these problems.

Background

The ability to support XML updates is a difficult and challenging problem because the basic structure (data model) underlying an XML document is a tree. The traditional (relational) database has a structure similar to a spreadsheet (a table with rows and columns) and the provision of table-based updates are a well-understood problem. However, updating a tree-based structure is much more challenging. There are four key differences between the simple table-based model and the more complex XML (tree-based) model.

The differences that set the XML data model apart from the table-based model present many unique challenges for the end-to-end management of XML documents, including storage, indexing, query planning, query processing, query optimization and result serialization. The provision of XML updates must facilitate all of these services in addition to supporting arbitrary node insertions and deletions within an XML document, and guaranteeing unique node identify and node order at all times. Indeed, the provision of efficient XML updates is still an open research problem. As the volume of XML data increases, and XML databases and XML repositories become more mainstream, the ability to support node-based XML updates efficiently becomes paramount.

Motivation

In my research, I analyzed several of the problems and bottlenecks in the provision of an XML update service and determined they were in fact symptoms of an underlying root cause – an inappropriate tree labeling scheme used in the encoding of the XML documents. Almost all existing research into dynamic labeling schemes for XML was developed in isolation. Each tree labeling scheme was designed to solve a particular problem or provide a particular feature, often without taking into account the impact their solution may have on other key features. To-date there has been no attempt to identify what exactly are the core properties (that are representative of the desirable characteristics) of a good and holistic dynamic labeling scheme for XML.

Contributions (to-date):

 

 

 

 

Other Research Interests:

The following are a list of some of the related areas in which I have an active research interest.

 

 

 

 

Journal Publications

Peer-Reviewed Publications

Master's Thesis (By Research)