Algorithm for geographical localization of criminals

A computer program is to help in the search for the mysterious sniper in maryland

A computer program is now to help in the identification and capture of the sniper who randomly shot seven people and wounded two others in the us state of maryland and the surrounding area. The rigel analyst software from environmental criminology research works on the principle of the "geographic profiling" and was developed by former canadian police officer kim rossmo, who is now director of research at the washington-based police foundation and is advising investigators in the present case.

algorithm for the geographic localization of criminals

Image: ecri

"The coordinates of each crime are entered into the system and the geographic pattern is then analyzed by an algorithm", explains rossmo. The system takes into account all available information about the perpetrator, such as the location of the crime, the sequence of events, evidence, and witness statements, such as a white van at the scene of the crime. The system deduces that the perpetrator is motorized and incorporates this information into the analysis. An algorithm then compares the data with the criminological expertise accumulated over decades on how serial offenders usually behave. Although the sniper may appear to be randomly and haphazardly shifting around, rossmo said, most traps usually follow a pattern that also fit this case and thus could provide a crucial clue to the perpetrator’s whereabouts.

A basic principle here is the "nearness principle". Research has shown that most single offenders commit their acts not far from their place of residence. "Criminals do not behave much different from normal people", according to rossmo. They generally stay in the terrain in which they can move with the means of transportation available to them. The software therefore takes into account in the analysis also distances that the perpetrator could cover when moving by foot, bicycle, bus or subway. However, there are exceptions: older perpetrators, in contrast to younger ones, sometimes accept longer distances to the crime scene, and women seem to travel longer distances for their crimes than dark-skinned americans. Normally, however, all perpetrators leave a buffer zone between the crime scene and their residential area in order to be able to act anonymously there.

algorithm for the geographical localization of criminals

Image: kim rossmo

This "hunting area" from the "residential area" is the task of the software to distinguish. As a result of the analysis, the police receive a geographical map of the case, which, according to rossmo, allows them to see to within 800 meters where the perpetrator might live, to which areas in particular further investigations should be extended accordingly, and which police databases need to be queried for specific clues, such as the observed perpetrator’s vehicle: "this allows an optimal search strategy".

The principle of geographic profiling is not new and has proven its efficiency many times before. For example, police in st. Louis were able to identify the "south side raptist", who had raped more than a dozen women over an eleven-year period, using this method to track him down by feeding the system with statements from victims and witnesses. Scotland yard, the fbi, the royal canadian mounted police and many other police stations around the world are also already working with the approximately 50.000 euro program. According to a study by geographic profiling experts of 450 crimes, rapes and murders in which the method had been used in the search for the perpetrator, in every second case the police were successfully helped by it. However, according to rossmo, even rigel cannot predict the location of the next crime. ()