?
Международный опыт применения математико-статистических алгоритмов прогнозирования преступности
The sphere of security provision is expanding and constantly bringing in new elements, including cyber- security, information security, computer network security, etc.). The arsenal of security tools is also grow- ing due to the ongoing proliferation of digital technologies (e.g. different technologies and telecommuni- cation channels for collecting, forming, processing, transmitting or receiving information related to security of the state). The article provides an analysis of current methods and technologies for crime forecasting in the national security domain. Achievements in the Data Science and Big Data generated the scientific basis for the development of Intellectual Data Analysis (Intellectual Analysis, Predictive Analysis), based on which mathematical and statistical forecasting of socially dangerous, criminal acts was designed (e.g. anti-terrorism algorithms, algorithms for predicting the activities of organized crime/gangs). The article aims to identify major trends and potential benefits of digital technologies pro- liferation as well as the challenges that states face while using mathematical and statistical methods for predicting crime. The meta-analysis of scientific researches and implementation of crime forecasting algorithms in different countries (such as USA, China, Japan, Singapore, India) helps to demonstrate a pluralism of approaches in the application of forecasting systems. The first part of the article presents the methodological and technical aspects of criminal data mining. The second part provides an overview of national practices in using crime prediction algorithms by the examples of Singapore, Japan, and India. The third and fourth parts are devoted to a more detailed analysis of the strategies and tactics of using algorithms in the USA and China, respectively. The analysis reveals the risks and benefits inherent in the most frequently applied mathematical and statistical crime forecasting algorithms. First, it is the “milita- rization” of the civilian sphere. Second, the algorithms, which do not take into account the social, cul- tural and political features of a given society, lead to the loss of statistical significance of forecasting. Third, historical data (recorded crimes) often contain racial, sexual, and contextual biases. Fourth, existing approaches do not pay heed to personal characteristics of a subject, as well as decision-making processes not infrequently resulting in wrongful conduct. Finally, there is no state control over the balance between the use of algorithms and respect for human rights.