Configuring the Matching Threshold

Using an arbitrary number is a way to determine from the data quality server whether the record being consolidated has a match in the Gold table or whether a new Gold record must be created. The number returned from the data quality software can be compared with the threshold values as per requirement.

The matching threshold configuration values for weightage and action in good and suspect matches determine how a consolidated record is treated and whether the record is created as a new entry in the Consolidation Gold object, linked to an existing entry in the Consolidation Gold object, or not processed. Records that are not processed are kept as a status of Cleansed or Unassigned so that they can be handled later by manual match.

You can set the following actions for Good, Suspect, and Below Suspect matches:

Process best match. If there is more than one match in the Consolidation Gold object, this option links the consolidation record to the match in the Consolidation Gold object with the highest weightage or score. However, in the case of a Multiple Table MASTER DETAIL Gold configuration, the best match is determined by considering the pattern first and then the score.

Process only if single match. This option links the consolidation record to the Gold record only if there is one match in the Consolidation Gold object. If there are multiple matches, it will not process the consolidation record, which can be handled later through a manual match.

Queue for manual matching. This option keeps the consolidation (staging) as cleansed, so that it can be handled later through a manual match. You can select the user group to notify in Notification. The records appear in the My Data Quality Inbox of every user in the Data Quality Steward Group. For details see Notifications.

Do not process. Available only for Below Suspect Match, this option is only applicable to records that do not achieve suspect match, at minimum. No action is performed on the consolidation record.

The best score among all candidate records is validated against all score thresholds, and the corresponding action is taken. For a multiple-table Gold Model, pattern takes precedence over score. First, the threshold pattern is evaluated, and then the score. The BOTH pattern indicates that the match score has been derived based on a match across attributes in both tables (Customer and Address tables in the customer MDM context), rather than just a match on attributes from only the master table (Customer table in the customer MDM context). Therefore, BOTH pattern is selected as the better pattern. For example, when there are two scores above the good score threshold (one with BOTH pattern and the other with MASTER pattern) the score with BOTH pattern is considered better than the score with MASTER pattern, even if the score of the BOTH pattern is lower.

For the Window Key Operator, when multiple window keys exist, the window key operator determines whether to use and AND or OR operator. For example, selecting OR to be applied between the multiple window keys would increase the range of potential matches.