The Use of Context-Dependent Modelling for the Construction of an Anti-fraud System in Transport

摘要

This article describes the results of applying a context-based approach to modelling when building a conditional system to counter fraud in transport. In this application, the exploration of the subject area into three contexts aimed at describing the characteristics of the driver's behaviour, his basic habits and the search for global anomalies based on historical data is considered The relationship and influence of context states is described by the system of the expert tree described. The description of the approach describes the advantages of the context-based approach when modelling analytical systems. At the end of the article, the results of testing the conditional system on synthetic data obtained from the collection of telematics data on the location and linear acceleration using a mobile device are described. It is shown that the proposed approach allows the detection of fraudulent actions of a telematics service user in more than 98% of cases with 16% of false positives.