Detection of large-scale fraudulent activities requires heavy usage of the online (real-time) data analysis. It requires complicated, and time-consuming investigations and deals with various domains of knowledge like financial, economics, business practices and law. Nevertheless, the real challenge is to build adaptive and self-learning fraud detection system, as it needs special methods of intelligent data analysis to detect and prevent losses.
On the one hand, desired fraud detection system must be able to deal with gargantuan computational complexity. It needs to recognise complex patterns over time periods spanning seconds to months. It also must be easily customizable and readily maintainable by specialists in frequently changing business environment. Ensuring compliance and finding fraud requires monitoring millions of daily transactions in real time and not error-prone invoice data which complicates processing and analysis. On the other hand, auditable proof of non-compliance is critical to tax enforcement and needs to be provided by the system too.
A state in Brazil dealt with revenue loss on VAT estimated at the level of 80M USD per year in retail only. Fraud Detection System created with the use of Cognitum Platform currently reduces lost tax revenues by 40%, and it is still learning. It now meals 2,000,000 transactions per day from 60,000 vendors at a speed of 200,000 rules per second.
The reasoning engine of the Fraud Detection System recognises complex fraud and non-compliance patterns. Natural language rules enable decision makers and specialists to manage and maintain tax fraud knowledge base by themselves, with an only sporadic support of programmers. Situation awareness is provided by operation dashboard that combines data visualisations and reports to give operators and management comprehensive situational awareness of fraud detection analysis and prevention.