EPSRC Funded Studentship: Efficient and scalable calculation of graph metrics for large-scale dynamic networks

A PhD position in network science is available in the groups of Drs Luc Berthouze and George Parisis in the Department of Informatics at the University of Sussex.

Graph processing is a key aspect of big data processing. An extremely wide range of physical and non-physical constructs can be modelled as graphs and therefore efficient graph processing and calculation of local and global metrics is crucial for a large and diverse set of applications. Computer, social, neuron and transportation networks are just a few representative examples of such constructs. Although increasingly sophisticated metrics are being developed to characterise networks, their applicability in the real world is severely limited by (i) the ever-increasing scale of the networks considered -- to the point that no single machine is capable of extracting basic graph metrics and (ii) the fact that most networks are dynamic, i.e., either or both of the nodes and edges change over time, and these changes can happen at timescales that can be as fast as the timescale of metric calculation.

To deal with the scalability challenge in static graph processing, a number of distributed graph processing systems have been proposed. These systems typically offer a generic, one-size-fits-all abstraction for operating on top of graphs that are partitioned across a cluster of servers. However, such approach does not cater to the fact that different domains involve networks with potentially very distinct structure and temporal dynamics. The aim of this PhD project is to develop new tools and methodologies for efficient and scalable calculation of key graph metrics for large-scale dynamic networks, via a two-pronged approach:

(a) designing, implementing and evaluating distributed graph processing algorithms for calculating graph metrics in an efficient and scalable way; (b) developing and implementing problem-specific graph metric approximations to tackle the challenge of fast-changing networks. This would be a key step towards network research having a more tangible impact in today’s Big Data.

University of Sussex is a research intensive university, in the sunniest part of UK, 50min by train to Central London, and 30min by cycle to the Brighton Beach.

-
Please apply for a PhD in Informatics via the University of Sussex postgraduate application system (http://www.sussex.ac.uk/study/apply). Include a brief statement of your scientific interests and skills/experience for the mandatory "research proposal" section, including how they relate to this project (maximum two pages). Indicate Dr Luc Berthouze as your preferred advisor and clearly state the title of the studentship in the finance section.

Good programming skills in at least one of Java, Python, Scala, C/C++ (essential)

Ability to work independently and be self-motivated (essential)

Experience with one of a number of distributed processing systems (e.g., Spark) and/or distributed graph processing systems (e.g., GraphX) (desirable)