Clique to host lecture by Dr. John O'Donovan: "WiGis: Web-based Interactive Graph Interfaces: A Scalable Approach"

Dr. Aaron Quigley, a co-Principal investigator on CLIQUE is pleased to host Dr. John O'Donovan in his CLIQUE seminar at UCD on "WiGis:  Web-based Interactive Graph Interfaces:  A Scalable Approach" in the UCD CASL seminar room at 11am on August 28th 2009.

Abstract

The abundance of network data available on social web sites such as Facebook highlights a need for dynamic and scalable network visualization tools, capable of meeting data-exploration requirements for a broad variety of users with different browsing devices and computational resources.   We believe that interactive network visualizations can be helpful in more complex tasks than simple data-exploration, in that they can be used to guide, control, and/or enhance openness and trust in complex peer-based processes such as product recommendation on Amazon.com or reputation modeling on eBay, for example.  To enable the application of interactive network visualizations to such problems, we require a visualization framework which is interactive, scalable, and easily accessible over the web, preferably in a single-click.   However, traditional network visualization tools are largely desktop-based, have poor interaction support and inherently suffer from scalability problems, especially when deployed over the web.
    The first part of this talk will focus on WiGis ña novel framework for Web-based Interactive Graph Visualizations.   I will explain why WiGis exemplify the first fully web-based framework for visualizing large-scale graphs in a user's browser at interactive frame rates.   This will include a live demonstration of interactive graph animations for up to hundreds of thousands of nodes in a browser through a novel use of asynchronous data and image transfer.   I will describe some comparative experiments in which our system outperforms traditional web-based graph visualization tools by at least an order of magnitude in terms of scalability, while maintaining fast, high-quality interactions.
    In the second part of the talk I will discuss ways in which the core framework has been applied to specific data-exploration problem in some example domains.   I will demonstrate scalable interactive visualizations of co-authorship networks based on research paper data, and an interactive visualization of the more complex process of collaborative recommendation, based on live data from Facebook.   In this example, I will explain how our WiGis visualization increases both transparency and trust in the recommendation process.    I hope to conclude the talk with your questions and an open discussion on the benefits, limitations and potential applications of the WiGis framework.
More information and live demos available at:  http://www.wigis.net

Speaker Biography

John O'Donovan is an Associate Specialist and Lecturer at the Department of Computer Science, University of California, Santa Barbara. Before this, John was a Postdoctoral Researcher at the same department. In 2008 John received his PhD from University College Dublin for his thesis entitled "Trust in the Social Web, Applications in Recommender Systems and Online Auctions" with Prof. Barry Smyth. This work was nominated by the School of Computer Science & Informatics for the 2008 national doctoral dissertation award. John also holds Masters and Bachelors (Hons.) degrees in Computer Science from UCD. During his PhD he spent one year as Visiting Researcher at the Viterbi School of Engineering, University of Southern California in Los Angeles.
John's research background is in Artificial Intelligence, with a specific focus on Trust Networks within Social Web Applications and other platforms for user-provided content. Like many other areas of study relating to the web, ideas from various disciplines are combined in his work. These include, but are not limited to: Data Mining, User Modeling, Network Visualization, Personalization, Natural Language Processing and Recommendation. His current work at UCSB focuses on Network Visualizations for the Social Web, specifically in ways they can be applied to recommendation and reputation systems. This work is funded by NSA, and has produced two US Patent submissions and two NIH Grant Applications.
Although a recent PhD graduate, John has had many collaborative colleagues and has co-authored over 20 international conference and journal papers, including a recent book chapter on Social Computing, published by Springer. John's most recent submission to SocialCom 09 received two best-paper nominations. His PhD work has been highly cited by the AI community. John has served on program committees and has reviewed for fourteen conferences and journals, and he is a member of AIAI, ECCAI and ACM.
In addition to his academic work, John regularly consults for EPS inc., a California-based software company specializing in intelligent web applications for the education domain. He has also consulted for Pacific Capital Bank, California Fire Department and three major US school districts.

Social Network Analysis Group

Search

User login

Subscribe to Clique

Sponsors