Using subscriber usage profile risk score to improve accuracy of telecommunication fraud detection

Look for telecommunications fraud in masses of call records is more difficult than finding a needle in a haystack. Even though fraud techniques are constantly evolving and whenever a detection method become know, fraudsters will adapt their strategies and try others. Therefore we propose a novel approach which make use of both threshold rules engine and usage pattern history for fraud detection in telecommunications. Threshold rules engine comprises of set of rules which if compromised would trigger an alarm. The alarm generated would be evaluated further through fraud risk score. The score is generated based on score value generated through subscriber's usage pattern on a formula formulated by domain expert. The alarm is categorized as fraud or vice versa depending on which range the scoring value falls into.