MDM Needs An Engine, Not Spark Plugs
By the Infoglide Team
Blog posts and marketing collateral often confuse identity resolution with data matching. When discussing master data management (MDM) solutions, you sometimes hear the term “identity resolution” applied when simple similarity matching is all that’s being used. That’s like calling a set of spark plugs an engine.
To create an engine that provides the power that MDM solutions need, you have to add five components to leverage great similarity matching algorithms.
First, the engine must orchestrate the actions of those algorithms so they provide confidence levels that define the degree of similarity among ambiguous data. But it also must distinguish relationships between different entities from identities. For example, you may have two records. One is for Mike Smith, 123 Maine Street, and one is for Michael Smithe, 122 Main St. You could reasonably assume these are the same person. But if the records also include date of birth, you could determine that since there is a 22 year difference in age between Mike and Michael, it’s likely that this is a parent-child relationship. So they’re not the same person but there is a relationship.
Second, the engine can enable the discovery of more complex networks of relationships (e.g., insider trading). Tracing relationships through multiple levels can reveal surprising connections, especially where people are working to hide their activities.
Third, these identities and relationship networks become the foundation for enhanced processing within the engine that determines the deeper meaning of the results by employing decisioning and rules processing to suggest - or take - action. So, let’s apply our example of Mike and Michael to the problem of insider trading. If Mike sits on the board of a company that is about to make a major acquisition announcement and Michael purchased stock in the company that is being acquired prior to that announcement, then that’s something that needs to be investigated.
Fourth, the engine needs to flexible. If it’s only targeted to a rigid structure, problem, or domain, making it solve a particular problem could range from expensive to impractical. So the engine needs to allow an organization to tune and adjust it to best meet the needs of that organization and the particular problem they’re solving. It can be targeted at problems ranging from insider trading to master person or master product indices.
Finally, this powerful engine is useless unless it’s easy to harness. The entire engine needs to be available to MDM solutions as a set of services that can be easily incorporated into new and existing business processes.
We’d be the last to say that a strong ability to match similar data isn’t important - yes even critical - to identity resolution. After all, your engine won’t run very long without good spark plugs.
However, MDM solutions that do more than just de-duplicate mailing lists should be supported by a complete identity resolution engine that is much more effective than simple matching algorithms in resolving multiple identities, uncovering hidden relationships, and applying domain specific rules to provide a recommended course of action.
Subscribe to this blog