A bank's credit card business development is mostly an off-system operation by business personnel, in which customer name, address and other information entry is off-site, leading to frequent issues of incomplete or incorrect input, inconsistent formats, and lower data accuracy compared to counter input, which leads to the need for a large number of subsequent manual data verification and confirmation work. Additionally, due to the upgrading of national anti-money laundering regulations and the continuous generation of new anti-money laundering alert rules within the bank, the validation rules for customer information gradually increase. As a result, the validation system needs to be passively upgraded, which requires a large amount of work and has a long response cycle. The main issues are as follows:
Aiming at the problem of irregularities in the data addresses of bank credit cards, Rock system adopts NER (Named Entity Recognition) address complementation strategy to complete the address information at all levels for complementation, reduces the model complexity by means of segment matching, strengthens the semantic understanding of machine learning algorithms, and applies entity recognition to identify cross-source customer information, so as to solve the problems of incomplete data, data errors, and different formats that occur when credit card information is entered off-site.To meet the ever-changing requirements for validating customer information, Rock offers a dynamic input of validation rules combined with existing rules. The validation rules are entered through the interface and take effect in real time, which shortens the cycle of new rules from being proposed to going live from weeks to minute level.
There are data errors and incomplete data in the billion-level stock data, and the retrieval speed is slow.
Manual validation requires a huge workload with low efficiency and poor results.
Traditional validation methods cannot meet all standards.
The new rule takes effect on a weekly basis.
Strong rule expression, covering all standards.
Unified data validation, one standard for all data.
Machine-executed validation greatly reduces the manpower required.
Enforce new rules in real time, immediately plugging the regulatory loopholes.
Real-time retrieval of massive data results, work no longer lags.
The accuracy of address error correction and completion has been improved to 97.7%.
The time required to verify a credit card customer's information is reduced from several tens of minutes to a few minutes, which greatly improves work efficiency.
Enforce new rules in real time, shortening the effective cycle of new rules from weeks to minutes.
Returns massive data retrieval results in real time, dramatically improving work efficiency.
Grandhoo