Remediating potential matches to improve data quality (IBM Match 360)

Review and evaluate potential record matches in your entity data. Make link or unlink decisions to improve data quality and increase confidence in your entity data.

Use the data quality capabilities of IBM Cloud Pak for Data and the IBM Match 360 service to identify, understand, escalate, correct, and prevent data quality issues within your entity data.

In this topic:

Before you begin
Viewing asset data quality and entity confidence
Generating tasks for potential match issues
Completing tasks to remediate potential matches

Before you begin

Before you can begin generating potential match tasks and remediating them, a data engineer user must have completed the following tasks:

Enable the clerical review range and configure clerical review and autolink thresholds. For more information, see Customizing and strengthening your matching algorithm.
Configure and activate the potential matches workflow for this service instance. For more information, see Configuring IBM Match 360 workflows.

Viewing asset data quality and entity confidence

Project owners and collaborators can view data quality information for their data assets. The data quality information is summarized in a series of dimension scores that you can access from your project's asset data quality tab.

One of the dimensions that contributes to the data quality assessment is entity confidence. The entity confidence dimension is only reported for connected IBM Match 360 entity data assets. The entity confidence score represents the percentage of master data entities in the system that IBM Match 360 is confident are correct and complete.

You can improve an asset's entity confidence score by tuning your matching algorithm or remediating potential match issues.

For more information about data quality, see Data quality information for data assets.

If your asset's entity confidence score is low, your data might benefit from generating and remediating some potential matches tasks.

To view data quality information and the entity confidence score:

From the Cloud Pak for Data navigation menu, select Projects > All projects. From the Projects page select your IBM Match 360 project.
Click the project's Assets tab, then select the connected master data asset for which you want to view data quality analysis.
Select the Data quality tab. It might take some time for the data quality report to generate and display.
When the data quality report is available, review the entity confidence score.

Tip: After data stewards have completed a significant amount of remediation tasks, return to the data quality tab to see how your asset's scores have improved.

Generating tasks for potential match issues

Before you can review and remediate potential issues within the matched entities in your data, you must generate some potential match tasks. You can either generate tasks from the full data set or from a set of search results to narrow the scope of the tasks. For example, you might choose to focus your remediation tasks on a certain geographical region to avoid overwhelming the task inboxes of your data stewards.

Required permissions: To create and remediate potential match tasks, you must be a member of the data engineer or data steward user group for the IBM Match 360 service.

To generate potential match remediation tasks:

From the navigation menu, click Master data explorer to open the master data explorer.
Optionally, run a search to narrow the scope of your data. For more information about searching, see Exploring master data entities and records in IBM Match 360.
Click Create tasks.
Define the parameters of your remediation task creation.
- Define whether you want to create tasks for all data or the results of your current search (if applicable).
- Select the types of entities to include in these tasks.
- Define how many tasks you want to create during this action.
Click Create tasks to start generating the remediation tasks.

When the tasks are ready, they appear as Remediate potential matches tasks in the task inboxes of assigned users. The user assignments are defined in the workflow configuration.

If an entity that was included in a generated Remediate potential matches task gets updated, either manually or through the matching process, Cloud Pak for Data automatically invalidates and removes the corresponding task. If there are still potential match issues after the matching update, you can recreate tasks by repeating the task generation procedure.

Completing tasks to remediate potential matches

As a data steward user, when you have potential match issues in your task inbox, you can claim them and remediate the identified data quality issues.

To remediate potential match issues:

From the Cloud Pak for Data navigation menu, choose Governance > Task inbox. Your task inbox opens to show your assigned or completed tasks.
Check to see if you have any Remediate potential matches tasks in your inbox.

Click the filter icon to access filter options you can use to narrow the scope of your inbox. For example, to show only remediation tasks: under Task type, select Remediation, then click Apply. You can also sort your task inbox by due date or date assigned.
To start working on a task, open it and click Claim task. When you claim a task, the system prevents it from being worked on by other users.
Review the summary details of the task.
Click Remediate issues to start comparing the records identified by the task. A graphical view opens, showing you a visual representation of all of the records and entities involved in this potential match task. Potential matches between records are represented as dotted lines.
Optionally, click each record to see its details.
In the potential matches panel, you'll see an item for each potential match decision included in this task. Click each item to review the details. You'll be shown a side-by-side comparison of the involved records to help you make your decision.

If you see a message saying Bulk action available, the system determined that it might be possible to make a bulk decision (either Link or Unlink) that applies to all of the issues in this task. Click Preview to see a visualization of what your changes will look like before making your decision.
Click Show entities to expand the comparison panel.
For each decision in the task, you can choose which action to take: manually link (Link), manually unlink (Unlink), or skip issue (Skip). If possible, the system will provide an intelligent recommendation, providing you with a confidence score for each recommendation.

Before making your final decision, you can also click Preview to see a visualization of what the entity and record structure will look like.

Make your decision, then confirm it in the modal by clicking Link, Unlink, or Skip.

Tip: The changes have not yet been applied to the data in your system. Click Submit to apply the changes.
Click Submit to apply your changes and complete this task. If you click Reject, then no changes are made.
From your task inbox, click Completed to see a list of all the tasks that you have completed, including details about each activity that you completed as part of the task.

Learn more

Parent topic: Exploring master data