The use of artificial intelligence algorithms in the criminalistic study of criminal activity (on the example of serial crimes)

Abstract

Since artificial intelligence technologies are firmly embedded in many areas of modern life, it is time to explore in detail their capabilities in criminology in the study of criminal activity. The article describes the experience of developing digital models of serial crimes committed for sexual reasons using artificial intelligence algorithms. The empirical basis of the study was data on 1068 serial crimes, including murder. Methods of mathematical statistics and a number of artificial intelligence algorithms were used for data processing: gradient boosting, neural networks, logistic regression, etc. As a result, typical signs of crimes and natural connections between them, used for training models, presented by artificial intelligence algorithms, are revealed. The obtained decision support systems allow predicting the distance from the crime scene to the place of residence of the offender (accuracy 88.3—93.5 %), the age of the offender (accuracy 80.3 %, confidence interval ±6 years), the presence of a mental illness (accuracy 81.5 %) and criminal record (accuracy 82 %), as well as some other characteristics. According to the results of the study, it is concluded that the methods of mathematical statistics and artificial intelligence can be used in the investigation of crimes, as well as in criminalistics.