Mastering Data Management

Probabilistic matching can help agencies achieve their MDM goals.

As government agencies seek to reduce the amount of improper payments – whether fraudulent or inadvertent – they are turning to master data management (MDM) to better link their data across the multitude of databases and agencies serving the public.

There are two primary types of algorithms used in most MDM matching: probabilistic and deterministic. Deterministic matching relies on exactly matching two datapoints to make an association.

Probabilistic matching systems, on the other hand, reduce error rates by using likelihood ratios, statistical theory and data analysis to accurately identify relationships between disparate, fragmented data, even data with complex typographical errors and error patterns.

Because probabilistic systems pinpoint variation and nuances to a much finer degree, they are able to identify associations that elude more traditional methods.

Agencies that apply existing statistical-based detection applications will be able to more accurately identify improper payments, reduce the need for manual review of suspect transactions and benefit from more accurate results.

In addition to ensuring that detection and analysis systems have the most complete and accurate information possible, MDM also provides an integrated view of large amounts of data in real time without requiring that data be moved into a centralized repository or hub.

MDM solutions that use a registry-based model enable agencies to easily and rapidly deploy solutions that match and link data from disparate systems without causing significant disruptions to their operations.

Registry-style solutions house data in each existing contributing source system or database. Data owners remain autonomous and retain control and responsibility for maintaining their data.

More importantly, a registry-style solution can enable data from hundreds of databases to be easily shared, even when data cannot be centralized because of regulatory, privacy or other business reasons.

Agencies seeking to solve their improper payments problem can achieve significant benefits from deploying advanced MDM solutions that are highly accurate, do not require that data be centralized and are capable of handling massive volumes of data in real time.

When the MDM solution feeds this significantly higher quality data into the detection systems already in place, agencies should be able to significantly reduce improper payment rates, which can translate into billions of dollars in cost savings annually for federal agencies alone, let alone state and local agencies.