Saturday, 12 March 2011

MDM Solutions and matching capability

Data Quality of the MDM Server is a very important aspect of MDM. The significant use case around Data Standardization, Matching, Suspect duplicate processing, are one of the key elements which differentiate the MDM capabilities from the Data Warehouse Solution.

Talking about the matching requirements of an MDM Solution, in any kind of a CDI solution it is very important to be able to hold of the unique identification of a customer/person. Consider an example wherein a person by name “Sushil Kumar” comes to a bank to open an account in the bank. This person has already defaulted credit card payment some 3 years ago and now because of may be difference in the spelling name or address change, the bank cannot make out that this is the same person. This could lead to wrong decision by the bank in allowing him to open a bank account again and providing him debit/credit cards. A similar error in health care can lead to a life and death situation. So an MDM Solution with good matching capabilities is required. The MDM Solution can either support matching capabilities OOTB or integrate with any of the other good matching engines.

Matching can be probabilistic or deterministic

Deterministic Matching
• Uses combination of algorithms and business rules to determine if two or more records match.
• The result is an either/or outcome: Either records match the requirements of the business rule or they don't.
• Better suited for businesses that have simple data systems with less attributes to match

Probabilistic Matching
• Uses likelihood ratio theory to assign comparison outcomes to the correct or more likely decision
• Probabilistic systems assign a percentage (such as 75 percent) indicating the probability of a match.
• Better suited for businesses that have complex data systems with multiple databases

IBM with its Quality Stage (QS) and Initiate matching engine now provides both the breeds of matching capabilities. The InfoSphere MDM Server has integration with QS to support matching while the IBM Initiate system has an inbuilt probabilistic matching engine. With time both the IBM MDM Solution can look at having support of both the matching capabilities and be able to incorporate any kind of matching.