Independent research paper finds that Minotaur reduces fraud losses by 40% over a 3 month period

The Journal of Scientific & Engineering Research has published a paper titled ‘Detecting Telecommunication Fraud using Neural Networks through Data Mining’, where Minotaur is recommended as providing a superior fraud detection rate to conventional systems, with 40% reduction in fraud losses and an 83% reduction in detection time over a three month period.

Date: 1st July 2012
Categories: Technology


Below is an extract from volume 3, issue 3. The full report can be read here: Detecting Telecommunication Fraud using Neural Networks through Data Mining.

How does Neural network help in reducing Telecommunications Fraud?

The Forum of International Irregular Network Access (FIINA) estimates that telecommunication fraud results in a loss of United States $55 billion per year worldwide (Turban, 2008, pp 349). South Africa’s largest telecom operator was losing over United States $37 million per year to fraud. Subscription fraud in which a customer either provides fraudulent details or gives valid details and then disappears was the company’s biggest cause of revenue leakage. By the time the telecom provider is alerted about the fraud, the fraudster has already moved to other target victims. Other types of fraud include phone card manipulation which involves tampering and cloning of phone cards. In a clip on fraud a fraudster clips on to customers telephone lines and then sells calls to overseas destination for a fraction of normal rates.

Minotaur, developed by Neural Technologies was implemented to prevent fraud. Minotaur uses a hybrid mixture of intelligent systems and traditional computing techniques to provide customer subscription and real time call monitoring fraud detection. It processes data from numerous fields such as event data records (switch/CDR, SS#7, IPDRs, PIN/authentication) and customer data (billing and payment, point of sale, provisioning) using a multi-stream analysis capacity. Frauds are detected on several levels such as on an individual basis using specific knowledge about the subscriber’s usage and on a global basis using generic knowledge about subscriber usage and known fraud patterns. The neural capability of Minotaur means it learns from experience making use of adaptive feedback to keep up to date with changing fraud patterns. In the first three months of installation of this neural network based software: The average fraud loss per case was reduced by 40%. and The detection time was reduced by 83%.. The combination of neural, rule based and case based technologies provide a fraud detection rate superior to that of conventional systems. Furthermore the multi stream analysis capability makes it extremely accurate.

Conclusion

The theft of telecommunication services has been one of the most enduring types of telecommunications crime which has been evident since the beginning of telephone systems (Grabosky P N and Smith R G, 2009, pp 84). Fraud detection for mobile telecommunications is a relatively recent area of research. Other works in the area of fraud detection in mobile telecommunications are based on data mining approaches (Althoff K D, 2008, pp 563).Due to its characteristics this type of fraud requires real time and individualized customer analysis. Each technological development designed to thwart criminal endeavors has been quickly followed by the creation of a new form of crime designed to exploit new security. These few directions of future policy may assist in ensuring that the full potential of global telecommunications developments will be realized while at the same time providing both service providers and users with some expectations that their property rights will be respected.

Fraud detection will continue their accelerated use of neural network based systems. Many of the laboratory techniques for using neural networks in metro burst communications, satellite network management and traffic control will come on line. In short the neural networks will become an increasing presence in major aspects of International Journal of Scientific & Engineering Research, Volume 3, Issue 3, March-2012 5 ISSN 2229-5518 IJSER © 2012 http://www.ijser.org

Telecommunication networks improving efficiency, adapting to changing calling patterns, and providing better information about the use of networks. Neural networks a technology which has been used in telephony since the early 1960s is beginning to make it presence felt in designing in telecommunications network of the next century. As a result Artificial Neural Network is a better method for detecting telephone fraud, due to its inherent ability to adapt as well as its speed and efficiency.

Press contact:

Top