PhD in Computer Science

Understanding cascades in complex networks with MANCaLog

Understanding cascades in complex networks with MANCaLog

Academic year 2013/2014

Sommario del corso

• Course objectives
• Notes
• Teaching tools

Course objectives

Social networks have become ubiquitous in modern life and offer unique opportunities for data mining.  In this talk, we look at challenges in mining information from social networks with multiple attributes and where geospatial information must be taken into account.

First, we introduce MANCaLog, a logic-programming framework for understanding cascades in complex networks.  We developed this framework by setting up several characteristics we believed such a language should achieve.  We then show how our framework meets these requirements yet still determine the outcome of a social network cascade in polynomial time.  This work was originally presented at AAMAS-13 and a follow-up paper will appear at ICLP-13.

Next, we look at the problem of finding geospatially disperse communities in a social network.  Previous work has shown that most community finding algorithms identify clusters of individuals that are geographically proximate.  This is often due to certain biases inherent in a social network.  In this work, we show how geospatial information can be discounted in order to find communities that are tightly-knit socially yet geographically disperse.  This work will be presented at KDD-13.

Finally, we present the Organizational, Relationship, and Contact Analyzer (ORCA) which uses many social network analysis and mining techniques to aide in police intelligence analysis for combatting gang violence.  We describe this piece of software and the challenges in integrating various algorithms in a single package as well as some preliminary results obtained working with our law enforcement partners.

Class schedule

Note

Paulo Shakarian is an Assistant Professor at the U.S. Military Academy (West Point, NY) in the Dept. of Electrical Engineering and Computer Science.  His research interests include logic programming, abductive inference, social network analysis/mining, applied artificial intelligence, and cyber-warfare.  His academic work has appeared in top conferences such as KDD, AAMAS, ESORICS, ASONAM, and ICLP.  His work has been featured in The Economist, WIRED, Popular Science, and Nature.  He also co-authored the new book Introduction to Cyber-Warfare: A Multidisciplinary Approach.  His website is http://shakarian.net.

• Teaching materials
• Exams and finals
• Go to Moodle
• Visit the forums