Product: Infosphere Master Data Management Advanced Edition
The probabilistic matching engine in MDM Advanced Edition provides a framework to enable leveraging multiple algorithms for your entity resolution needs.
In the party domain, this capability allows MDM to programmatically choose one deduplication algorithm and zero-to-many other search algorithms, for a given person or organization.
You can further extend this default behaviour to essentially define multiple deduplication algorithms for the person and organization domains. Your custom implementation determines a single deduplication algorithm, among those deployed, that is applicable for any given person or organization party record. This essentially supports implementations where dedicated algorithms are desired per various classes of Person or Organization records, as long as a single person or organization record is associated to only one of those classes.
Here are some usecase examples for this capability:
1- You may implement your MDM to leverage country-based deduplication algorithms for the parties residing on those countries.
2- You may deduplicate prospect and lead parties, separately from customers in your MDM.
3- You may deduplicate your customer parties within their lines of business.
4- You may deduplicate your party records based on a strict algorithm (to avoid false positive cases), while a secondary less-strict algorithm to assist the data steward resolutions.
5- You may algorithmically assign your party records into different group types (like householding)
For additional information, here’s a link to the knowledge center in regards to suspect duplicate processing using PME https://www.ibm.com/support/knowledgecenter/en/SSWSR9_11.6.0/com.ibm.mdmhs.dev.party.doc/concepts/c_understandingprobabilisticmatching.html#MDM