Attention MDM Advanced Edition clients!
Machine Learning assisted Data Stewardship is currently only available for virtual MDM (Standard Edition). In order to make similar functionality available for physical MDM (Advanced Edition) we are looking for MDM AE customers that are willing to share some (strictly non-PII) training data from their MDM systems with MDM Development that would be used as training data for machine learning models, to establish that the same or similar approaches are valid for physical MDM as well.
We would also appreciate for any customer to review early UX prototypes of a Machine Learning assisted Data Stewardship User Experience.
Steps involved generating the training data:
To generate the training data, you would need to execute a pre-release version of the extraction tool that is shipped with MDM. This pre-release version contains the required steps for AE data extraction. Customers need to run this tool from a command-line on the MDM instance.
Content of the training data:
The extraction tool generates a CSV file containing the training data. Each line in the CSV file represents a decision of a data steward on a suspected duplicate. The data contains the decision of the steward as well as comparison details of the two persons the decision was taken on. These comparison details are numerical values provided by the matching engine and do not contain any information about the person records themselves, i.e. no PII.
Example of the training data:
Senior Manager MDM Development - MDM Program Management, Machine Learning and Consent Management
IBM Research & Development GmbH