Investigative Graph Search using Graph Databases

Abstract

Identification and tracking of individuals or groups perpetrating latent or emergent behaviors are significant in home-land security, cyber security, behavioral health, and consumer analytics. Graphs provide an effective formal mechanism to capture the relationships among individuals of interest as well as their behavior patterns. Graph databases, developed recently, serve as convenient data stores for such complex graphs and allow efficient retrievals via high-level libraries and the ability to implement custom queries. We introduce PINGS (Procedures for Investigative Graph Search) a graph database library of procedures for investigative search. We develop an inexact graph pattern matching technique and scoring mechanism within the database as custom procedures to identify latent behavioral patterns of individuals. It addresses, among other things, sub-graph isomorphism, an NP-hard problem, via an investigative search in graph databases. We demonstrate the capability of detecting such individuals and groups meeting query criteria using two data sets, a synthetically generated radicalization dataset and a publicly available crime dataset.