Yekta, an integrated online fraud detection system, can be installed alongside any system where there is possibility of fraud in it. By data mining and modeling the behavior of significant entities in that system, Yekta detects and reports suspicious and unusual activities and transactions.
Due to benefiting from artificial intelligence and the new level of technology used in it not only Yekta system does not hinder valid activities of customers, but also in the face of new violations or changes in customer behaviors, it becomes more powerful and efficient over time.
Once installed alongside card payment system in a bank or financial and credit institution, at first, entities such as customers, POS devices, ATMs, and various terminals are identified. Then, with the help of special AI and Machine Learning algorithms that are very complicated and yet fast, the performed transactions are intellectually checked online. The degree of suspicion of fraud is determined and necessary warnings are announced in the system output channels. Finally, depending on the policies of the organization, either the transaction is prevented or other appropriate actions are taken to verify the authenticity of the customer and the transaction.
The Yekta system can consider all the filters that the organization has logically or experimentally obtained, without preventing valid transactions.
Pooya Online Fraud Detection System (Yekta) is an innovative and knowledge-based system. Relying on 38 years of successful experience in providing the latest automation solutions in banking industry, Pooya has designed and developed this system in cooperation with a research team of elite professors and university specialists, expert in the field of AI. Yekta is at the service of Iranian banking needs in a completely localized and efficient way.
- High detection rate of suspicious transactions and close to zero rate of hindering valid transactions
- Smart filtering of highly valid transactions
- Deploying AI and Deep Learning algorithms
- Capability of detecting new patterns of fraud as well as recognition of fraud in case of changes in the patterns, by relying on learning power
- Benefiting from parallel processing and distributed processing methods
- Ability of online response by using pre-stored algorithms for fast recognition
- Ability of updating fraud detection models, periodically
- Capability of modeling behaviors (user, device, …)
- Extracting more than hundred behavior specialties of individuals and payment channels, per transaction
- Detecting hidden behaviors as well as suspicious trends
- Creation of a live and growing system with capabilities of self-correction and performance improvement, over time
- Online and offline separate sections
- Modeling of customers behavior patterns, offline at nights, when there are less transactions performing
- Comparing online function of a customer with his personal behavior model, and recognition of percentage of deviation from his normal behavior
- Comparing the deviation of customer action with the fraud threshold and reporting the result to the system
- Taking into consideration, the logs of customer transactions and payment channels, as a chain of information
- Supervising confirmation of transaction validation as well as confirmation of customer authentication, and hindering unconfirmed transactions
- Elimination of human error in traditional methods of inspection
- Insertion of required information in the system database and generating various reports
- Increase of customer satisfaction
- Supporting ISO 8583 communication protocol
Failure of organizations to take appropriate action to prevent fraudulent attacks not only increases the risk of large financial losses to them and their customers, but also undermines their prestige and credibility in society. Moreover, it imposes additional costs on judicial and disciplinary systems of the country for handling such cases.
Entering of Artificial Intelligence (AI) to the field of fraud detection
Although filtering can prevent criminals from influence in a period of time, but is it effective against new methods of fraud and changes in customer behaviors too? In order to a fraud detection system can automatically improve itself against changes in the ecosystem of application field over time and yet does not lose accuracy, AI specialists rushed to the aid of this front to fight the criminals. Today, products in this field are compared and evaluated based on the extent of use of the highly accepted methods of artificial intelligence.
A high percentage of banking frauds and violations in Iran is related to abusing people’s bank card information. The methods such as card theft, card skimming, CNP frauds, phishing, card copying, and fake transaction are committed.