* Faculty       * Staff       * Students & Alumni       * Committees       * Contact       * Institute Directory
* Undergraduate Program       * Graduate Program       * Courses       * Institute Catalog      
* Undergraduate       * Graduate       * Institute Admissions: Undergraduate | Graduate      
* Colloquia       * Seminars       * News       * Events       * Institute Events      
* Overview       * Lab Manual       * Institute Computing      
No Menu Selected

* News


Mining Signed Information Networks with Applications to Detecting Fraud and Polarization

Speaker: Leman Akoglu
Stony Brook University

November 4, 2013 - 11:00 a.m. to 12:00 p.m.
Location: JEC 3117
Hosted By: Dr. Sibel Adali (x8047)


Graph mining has been a growing area of research in the past few decades. Researchers have studied various forms of graphs such as time-evolving graphs, disk-resident graphs, and attributed graphs and their associated algorithmic problems. Recently, signed graphs have also attracted attention with problems related to edge sign prediction, where signs depict the trust/distrust, up/down vote, friend/foe relations between the same type of entities. In our work we study (negative/positive) opinions of individuals, often exerted in online media, which we represent as signed bipartite graphs between the individuals and other types of entities. Exploiting the "network effects" in these signed graphs, we target two seemingly unrelated problems: (1) spotting fraudsters and fake reviews in online review systems, and (2) identifying and ranking political polarity in debate platforms. In this talk, I will introduce the opinion fraud and political polarity detection problems and describe a fast, effective, and extensible algorithmic framework to address these problems in a unified fashion. Our method exploits only the signed network structure unlike many works that solely focus on text information in opinion mining. I will discuss results on real-world datasets including the AppStore andPoliticalForum.com, as well as several future directions to integrate side information to the proposed framework for better detection accuracy.


Leman Akoglu is an Assistant Professor at Stony Brook University. She received her PhD from Carnegie Mellon University in 2012 and her BSc from Bilkent University, Turkey. Her research interests span a wide range of data mining and machine learning topics with a focus on algorithmic problems arising in graph mining, social networks, and anomaly and event detection. Leman's research has won 2 Best Paper awards and led to 3 U.S. patents filed by IBM T. J. Watson Research Labs.

Last updated: October 15, 2013