Share via

Task 2: Testing and Publishing the Matching Policy

  • Article

In this task, you test and publish the Remove Duplicate Suppliers matching policy.

  1. In the Matching Results page, click Start to test the entire policy. In your case, you have only rule in the policy, so the results from testing the rule and the policy should be the same.

  2. Review all the matched records and their matching score in the list box. A record that has a Green icon associated with it is a duplicate of the pivot record that precedes it. Here are couple of examples:

    1. The record with Record ID: 1000005 is a match of the record with Record Id: 1000004 with Score: 100% because both the records have the same values for SupplierID (prerequisite), Supplier Name, and ContactEmailAddress columns. DQS randomly picks a record as the pivot record for a cluster.

    2. The record 1000023 is a match of the record 1000022 with the matching score: 93% because the two records have the same values for SupplierID (prerequisite) and Supplier Name columns, but different values for the ContactEmailAddress column.

    3. Scroll to the bottom of the list to see two records with records IDs: 1000051 and 1000052. Record 1000052 is considered a match with matching score 91% because the two records have the same values for the SupplierID and ContactEmailAddress columns, but different values for the Supplier Name column.

    Policy Definition - Policy Results

  3. Right-click on any matched record (with green icon) and click View Details to see more details about the matching such as contribution of each field score to the overall matching score.

    Matching Score Details Dialog Box

  4. Click Close to close the Matching Score Details dialog box.

  5. Click Matching Results tab at the bottom of the page. This tab gives you detail such as number of matched records, number of unmatched records, number of clusters with matched records, the average cluster size, minimum cluster size, and maximum cluster size. See Create a Matching Policy for more details. You cannot export results from this activity. You are just defining a matching policy by using the sample data to test rules and the policy against the sample data.

    Matching Results Tab

  6. Click Finish to finish creating the matching policy.

    Note

    You have defined the matching policy here; therefore you cannot export results to an output file. You basically used a sample input file, created rules, and tested the rules and policy against the sample data with the goal of defining the policy.

  7. In the SQL Server Data Quality Services dialog box, click Publish and click OK on the message box. Now, the matching policy you defined is published into the Suppliers Knowledge Base. You can use the knowledge base to run the matching process against an input file to identify and remove duplicates.

Next Step

Task 3: Creating and Running a Data Quality Project for Matching