Data Mining of Forensic Association Rules

Data mining offers a potentially powerful method for analyzing the large data sets that are typically found in forensic computing (FC) investigations to discover useful and previously unknown patterns within the data. The contribution of this paper is an innovative and rigorous data mining methodology that enables effective search of large volumes of complex data to discover offender profiles. These profiles are based on association rules, which are computationally sound, flexible, easily interpreted, and provide a ready set of data for refinement via predictive models. Methodology incorporates link analysis and creation of predictive models based on association rule input.