Summary
Within the Match and Merge solution, the matching algorithm scores potential duplicate records based on static match criteria. Using this score, the solution auto-merges duplicate records above an upper threshold, rejects duplicates falling below a lower threshold, and places all records falling between those thresholds into clerical review. The number of clerical review tasks can often be substantial.
Now, the Machine Learning Match Recommendation (MLMR) works alongside the daily activities of the data steward by observing their duplicate review decisions during clerical review sessions and continuously refines its learning. As the MLMR reaches sufficient certainty, it will begin to provide its own match recommendations, which the data steward can use to make decisions more quickly. These recommendations result in a rapid depletion of the clerical review backlog, saving the data steward significant time.
Note: The functionality for this release is in the ramp-up phase and is only accessible through an early adopter program. To learn more about the ramp-up phase / status, refer to the License and Component Lifecycle topic in the System Update and Patch Notes section of online help here. To participate in the early adopters program, send an email request to SYSClericalReviewMatchingAgent@StiboSystems.com.
Details
The Machine Learning Match Recommendation is non-invasive, meaning that it works in tandem with the daily activities of the data steward by continuously observing and learning from the skilled match decisions they make when working with clerical review task lists. It then provides its own merge or reject recommendations once it has the confidence to do so. Over time, more items in the duplicate review backlog are covered by its recommendations with a higher degree of accuracy.
The Machine Learning Match Recommendation helps the data steward to maintain complete control. It provides a recommendation to the data steward who can then act on each recommendation individually.
For additional information about the challenges CMDM customers face and the solution to those challenges that the machine learning-based matching agent represents, click on the video below. If it does not play as expected, it is also available in the Customer / Partner Communities and may also be accessible within Stibo Systems Service Portal.
For more information about this functionality, refer to the Machine Learning Match Recommendations topic in the Matching, Linking, and Merging documentation here.