The keynote presentation at the 2nd Jordan International Conference on Computer Science and Engineering discusses three examples of modeling complex networks using computer science techniques:
1) Analyzing the link structure of websites using link structure graphs to understand site organization and user experience.
2) Studying messages exchanged in online social networks using machine learning to determine levels of agreement and disagreement.
3) Modeling complex biological systems like gene networks using stochastic pi-calculus to study multi-component system interactions.
The presenter is Dr. Natasa Milic-Frayling, a senior researcher at Microsoft Research Cambridge who leads research on information retrieval, machine learning, and user-centered design.
1. Keynote Presentation at JICCSE
The 2nd Jordan International Conference on Computer Science and
Engineering
Dec 5, 2006
Title:
Modelling and Analyzing Complex Networks
Abstract
From the Web content structures to online social interactions and intricate biological systems,
we find opportunities for applying computer science techniques to expand our knowledge and
gain new insights in the network phenomena. In this presentation we provide three examples
of modelling structures, interactions, and processes associated with networks: the analysis of
web site organizations using Link Structure Graphs, the study of messages exchanged in
online social networks using machine learning techniques, and modelling of complex
biological systems using the stochastic pi-calculus.
Organization and evolution of the Internet has been modelled and studied using a Web graph
which comprises nodes that correspond to Web pages and links that represent connecting
hyperlinks. However, understanding the structure of individual Web sites and capturing the
properties that arise from the organization of hyperlinks on individual pages requires
alternative representations. We demonstrate how a novel Link Structure Graph technique
allows us to investigate site properties and how these properties affect the user perception of
the site and the browsing experience.
Internet has unleashed new opportunities for socializing through online communication. Users
form discussion groups where they debate issues or ask for help and get answers to their
questions. In many application scenarios it is important to characterize the dynamics in the
participants’ interaction and the roles that individual participants play. For example, heated
discussions can sometime end in outburst of disagreement and abusive language and one
may want to avoid reading such content or joining the group. We demonstrate how machine
learning techniques can be applied to exploit properties of the discussion group network and
automatically determine the level of agreement and disagreement in a particular discussion
thread.
Finally, studying complex biological systems, such as a gene network, presents great
challenges for scientists. A common approach is to study functions of individual proteins, thus
falling short of explaining the functioning of the system they form. Fortunately, the stochastic
pi-calculus offers a breakthrough in this area. Originally developed for modelling concurrent
systems in Computer Science, models actions and processes and offers an effective way of
studying multi-component systems. Applied to the gene network it enables modelling and
studying of complex gene interactions.
2. Presenter’s Biography:
Dr. Natasa Milic-Frayling
http://research.microsoft.com/users/natasamf
Senior Researcher
Microsoft Research Cambridge
http://research.microsoft.com/aboutmsr/labs/cambridge/default.aspx
United Kingdom
Natasa Milic-Frayling joined Microsoft Research Cambridge, UK in 1998. She has been
setting research directions for the Integrated Systems group. Her interests include design and
evaluation of information retrieval systems, machine learning methods for text, user centred
design of information management systems, and technologies for cross-platform and context
rich communication.
Since October 2004, Natasa has been heading the Research Partnership Programme that
facilitates collaboration between researchers and Microsoft teams in MS Subsidiaries across
Europe, Middle East, and Africa (EMEA).
Natasa obtained her B.S. in Applied Mathematics from University of Zagreb, Croatia in 1984
and Ph.D. in Applied Mathematics from Carnegie Mellon University, Pittsburgh, PA in 1988.
Prior to joining Microsoft Research, she worked at Claritech Corporation (currently
Clairvoyance Corporation), a spin-off company from Carnegie-Mellon University, producing
software components for building information management systems. With the acquisition of
Claritech by the Justsystem of Japan, Natasa assumed the role of the Director of Research.
Natasa is actively involved with a wider research and academic community. She is publishing
at academic conferences, participating in program committees of academic events, and
promoting research and innovation through public and academic presentations.