Since 2013 I have been involved in collaborative research efforts to detect cyberbullying in social networks. Collaborators include Dr. T. Solorio at the University of Houston and her student Suraj Maharjan and Dr. R. Diaz-Sprague, Visiting Scholar at UAB. The goal of this research is to explore technological ways to reliably detect invective in social network communications and possibly nudge users toward wisdom.
Applying Data Mining to Internet Security
We cluster spam email messages in order to gain an understanding
of the world of spam. The largest spam campaigns commonly show up
as the largest clusters.
Sometimes our methodology is to cluster the spammers' hosting domains,
rather than clustering the emails themselves.
One goal is to enable law enforcement to focus on the largest spammers.
Spam Data Mine
We are also working on phishing and malware. We are using clustering
to study phishing, and closed frequent sets to study malware.
Center for Information Assurance and
Joint Forensics Research
Fundamental Algorithms in Data Mining
KDDM Lab web page
Algorithms to compute frequent itemsets, especially closed
frequent itemsets; use of network flow to discover outliers.
Algorithms for special classes of graphs.
The emphasis here is on efficiency. Often, this involves
exploring the "border" between polynomial time problems and
Inferring a context free grammar for a domain specific language,
given (positive) samples of the language.
Ph.D. Computer Science, Ohio State University, 1988
Advisor: Ten-Hwang Lai.
Ph.D. Mathematics, Ohio State University, 1973
Advisor: Dijen K. Ray-Chaudhuri.
Finite State Automaton (Nondeterministic) Simulator