You are here: Home Research Research Interests: Sonia Berman
Site Options
Personal tools
Document Actions

Research Interests: Sonia Berman

Prof Berman is looking for PhD and MSc students in the areas below.

XML Databases The goal of this research is to facilitate the integration of two disparate kinds of information, namely databases and semi-structured XML documents, so that users can access both kinds of information source in simple, flexible and efficient ways. Effective techniques for storing, querying, updating, mapping and distributing XML and hybrid data systems need investigation. One MSc student in the group is currently working on temporal XML databases, another on querying of hybrid systems; a number of related issues are open for study.

Peer-to-peer Database Systems In a peer-to-peer system a number of autonomous databases join together, for different intervals of time, in a system which has no global schema and no central administration authority, yet is usable as if it were a single store. Aspects to investigate include network topologies, clustering and routing policies, query evaluation and issues of trust in P2P systems. A current MSc student in the group has built a prototype peer-to-peer database system which can be used as a basis for experiments in this field.

Advanced Database Applications Suitable models and tools are essential for databases to be easily and effectively exploited in complex application systems. Some examples are given below.

  • As the amount of online information rapidly escalates, we need to provide better knowledge management systems to organize this more effectively and make finding and sharing information simple and efficient. Knowledge sharing by trusted individuals with common interests should be an important benefit of the Web, but we have away to go to ensure this is done using appropriate structures, tools and collaboration approaches
  • With the increasing use of personal information devices, we need to address the issue of how to provide an information-centric experience for the end user across devices. Code deployment, personal databases, and the ability to update replicated information from many devices while connected or disconnected, poses challenges for applications running on the small devices and also for central management of such systems.
  • In a GIS (Geographic Information System) the spatial and temporal nature of data affords new challenges as regards modelling, design, storage, visualization and analysis of large data sets. Research in this field can range from engineering methodologies for GIS design through to indexing structures for spatio-temporal data. Recent MSc graduates, and some current thesis students, have developed models and tools to facilitate working correctly with maps and related GIS data, and we have several partner organizations involved in this work with us.

Web-based Information Systems Database research is now being applied to the World Wide Web in order to improve information access and management for this enormous and rapidly growing resource. Data management in this field covers both the Surface Web and the Deep or Hidden Web, which includes over 25 million databases for which query forms exist on the Web. Better approaches are needed for locating, integrating, querying and personalising access to this vast store of information.

Object Store Engineering Research in this area concentrates on developing better techniques for object databases/object stores. Since the structure and use of such stores is very different from that of the simpler relational model, new mechanisms and approaches are needed to handle distribution, store organization, query optimization, compile-time optimization, code in databases, class evolution, etc. Recent MSc graduates produced an object store and caching mechanism which can be built upon if you wish. If you are interested in working at this level, another avenue of research is main memory
database systems. Main memory can now be so large that accommodating entire structures such as indexes or database partitions in memory is a viable option. This requires very different approaches to indexing, caching, store organisation, etc. which have yet to be explored.

Note: Prof Berman is willing to look at other ideas for post-graduate work in the database field if you have any specific interests.

last modified 2007-07-31 08:35