Using the AI Model

After the AI model has been deployed, use it to match incoming records with the existing records. The AI model gives accurate matching results during the matching process, which in turn helps you to make the right decisions when merging records.

Before you begin See Managing the AI Model.
    Procedure
  1. In ibi MDM, navigate to Business Processes > Process Definition Selection > Default > Create.
  2. On the New Rule page, from the Set the Process to list select the generated data quality workflow. See Business Processes.
    These workflows contain newly generated Java classes. These classes are used by the Matcher activity. Whenever a new record arrives in ibi MDM, these classes are used.

  3. Set Match Record Approval to the newly generated MatchAttrList_repository name.xml file.

  4. Add a new record to the primary repository and the child records to the child repositories. see Creating Records.
  5. View event details for the latest Record Add event. See View Event Details.
    The Status of the Matcher process must be Success. If the MatchCount is more than 1, the model has found a number of duplicates and a work item is created in the Inbox.
  6. Navigate to Inbox and open the new work item.

  7. Merge the incoming record with one or more found records, or accept the incoming record as a new record.

ResultYou must make merging decisions carefully and follow the same business principles because these decisions are used to further train the model. Randomly selecting records to merge, or skipping merge action for records that actually match the incoming record decreases the model accuracy. You can also find and merge several existing matching records in the primary repository using record UI. These merge decisions are also used to further train the model.
What to do nextSee Retraining and Redploying the AI Model.