Extending automated intelligence systems via graph database: A case study of the "Meth Hunter"

Abstract

Automated intelligence analysis was born into a skeptical community dealing with high stakes dilemmas, where often outcomes are measured in loss of human life. The massive amount of data require analysts to be masters of identifying "indicators" intuitively foreshadowing intelligence targets and masters of modeling the indicators in analysis systems. The paper explores the application of graph theory through graph databases to determine how the index-free adjacency of graph databases can be exploited to improve processing time. The study utilized the Meth Hunter, an analytic machine designed to identify methamphetamine conspirators by analyzing pseudoephedrine purchase records as a case study. The graph database (Neo4j) was compared to a SQL relational database (WAMP). Neo4j demonstrated superior performance in identifying and retrieving relationships between data points. Neo4j successfully demonstrated the ability to implement such a strategy and extend the horizons of traditional data mining systems.