Automated Crime Profiling
". . . experts in criminal personality profiling are taking advantage of the existing technology of artificial intelligence . . . to capture the elusive decision making rules associated with the profiling of serial violent criminals."
By
DAVID J. ICOVE, Ph.D., P.E.
Senior Systems Analyst
Behavioral Science Investigative Support Unit
National Center for the Analysis of Violent Crime
FBI Academy
Quantico, VA
In the fall of 1983, Special Agents from the FBI's Behavioral Science Unit constructed a criminal personality profile describing an individual who could have been responsible for a series of fires at religious homes and houses of worship that summer in a posh New England community. The profile was prepared at the request of the community's police department, which later discovered that the FBI's profile not only accurately described the suspect in detail but also pinpointed his residence, based upon a series of intricate computer calculations using artificial intelligence technology. The suspect later confessed to the crimes.
This pioneering use of artificial intelligence technology in crime analysis and criminal personality profiling provided the groundwork for the present automation efforts at the FBI's National Center for the Analysis of Violent Crime (NCAVC).1 In an active project at the NCAVC, experts in criminal personality profiling are taking advantage of the existing technology of artificial intelligence, or AI as it is known to its users, to capture the elusive decision making rules associated with the profiling of serial violent criminals.
This article is an overview of the ongoing research and development efforts by the NCAVC to automate the criminal personality profiling process. Future articles are planned to advise law enforcement researchers and investigators as to the success of this exciting and thought-provoking technology.
Violent Crime Model
The relationship of AI to the profiling process is best described using the "Violent Crime Systems Analysis Model." (See fig. 1.) This model was developed during the conceptualization and development of the NCAVC computer systems and traces the philosophical activities involved with the detection, prediction, and prevention of violent crime.
The model is divided into reactive and proactive investigative strategies. Reactive strategies include crime scene investigative support during immediate response to incidents, while proactive strategies explore effective anti-crime programs to both deter and apprehend offenders.
Briefly, the model emphasizes the reporting of violent crimes (step 1) to the NCAVC for crime pattern analysis and classification. This information may come from written media accounts (step 2), crime scene processing (step 3), VICAP crime reports (step 4), or violent crime research findings (step 5). Crime pattern analysis (step 6) can determine if any case trends are detected in the profiled incident that have existed in the past, predict the probability of the occurrence of future incidents, and check for the possible identification of prior known/unknown criminal offenders based upon their past methods of operation.
Crime pattern recognition analysis can also classify incidents into naturally occurring groups, such as the type of
