Behavior Signatures

MYMIC LLC, working with the University of Alabama, Huntsville, proposes a methodology to use behavioral signatures to predict threat actions. The technical approach we propose to use to identify behavior signatures for and predict terror attacks is Bayesian data analysis. This methodology is a collection of practical statistical methods for making inferences from data, based on probability models developed from that data for both observed and unknown quantities of interest. The proposed behavior signature development and prediction methodology is built on two key ideas: a database of past attacks which will be analyzed, and the Bayesian statistical methods for analyzing the database and calculating the probabilities of future attacks. The goal is to analyze the attribute data in the database for predictive patterns, such as: 1.0 Temporal attributes (e.g., time of day, proximity to holidays or political events) 2.0 Causal linkages (e.g., retributions and re-retributions for prior events) 3.0 Location or locale (e.g., city A more likely than B, or terrain type X is more likely than Y)