Creating, Editing, and Deleting Matching Rules

Matching rules control how nodes that are named differently are matched when matching request items or deduplicating a viewpoint. Instead of an exact match on the node name, you can perform matches based on other properties and match types.

Matching rules are created at the node type level. You can define multiple rules for a node type, and you can specify the order in which the rules are evaluated. You can also specify thresholds to automatically accept or exclude match results based on their confidence match.

Matching Rules Overview

When matching request items for new nodes in a request or when matching existing nodes in a viewpoint, by default nodes are considered to be the same node when there is an exact match name on the Name property (including any node type qualifiers and alternate names). Matching rules enable you to specify other properties and match types to determine whether nodes are the same.

To illustrate the matching concept, let's look at an example of matching request items. Suppose you had a viewpoint with a node named "Oracle Corporation", and you upload a request file that contains a node with the name "Oracle Inc". If you were to apply the changes from this request file without matching, you would end up with two different records (because there isn't an exact match on the name).

Instead of the exact name match, you could set up a matching rule that looks for names that are similar to one another. Then, when you run matching on the request items, the system presents the incoming "Oracle Inc" node as a potential match to the existing "Oracle Corporation" node. When the match is accepted, the incoming node is merged into the existing node, and you have a single record for Oracle with the merged information from both nodes.

You could also configure the matching rule to match on properties other than Name. Suppose both records had a property called Symbol, with the value "ORCL". You could set up a matching rule to match on Symbol instead of Name in order to match the incoming node to the existing one.

Best Practices

While you can enable up to three match rules at a time, each rule requires additional processing time when running the matches. Therefore, it is a best practice to enable only the minimum number of matching rules that meet your business needs.

Considerations

  • You must have Owner or Metadata Manager permission on the application or dimension that contains the node type in order to create, edit, or delete matching rules.

  • If you enable multiple matching rules, matching is performed using OR logic. That is, if a node meets the criteria of one rule or the other rule, it is presented as a potential match.
  • If you add multiple criteria to a single matching rule, matching is performed using AND logic. That is, in order for the nodes to be considered a match for a rule, all of the criteria that you specify must be true.
  • Matching rules for registered data sources can be used for either matching request items or deduplication. Matching rules for unregistered data sources can be used for matching request items only.

Creating Matching Rules

  1. Inspect the node type that you want to create a matching rule for. See Inspecting a Node Type.
  2. Navigate to the Rules tab, and on the Matching sub-tab click Create.
  3. Enter a name and, optionally, a description for the matching rule.
  4. In the Data Sources drop down menu, select the data source that will use this matching rule, and then click Create.

    The matching rule is displayed in the inspector.

Editing Matching Rules

After creating a matching rule, you can edit it to change the name or description, enable it, change the rule order, set thresholds to automatically accept or exclude a match result, or edit the definition parameters. You cannot edit the data source for a matching rule after it has been created.

  1. Inspect the matching rule that you want to edit.
  2. On the General tab, click Edit and perform an action:
    • (Optional) Edit the name or description.
    • In Rule Order, use the Up and Down arrows or enter an integer to specify the order in which this matching rule should be run for this data source and node type.

      Tip:

      It can be helpful to enter larger values for the rule order so that if you create new rules that you want to place before your existing rules you do not have to manually reorder all of your rules.
    • Use the Enabled checkbox to enable or disable the matching rule.
    • In Usage, specify whether the rule will be used for Matching (see Understanding Matching and Merging Request Items) Deduplication (see Understanding Deduplication), or both.

      Note:

      You can specify a Usage for matching rules for registered data sources only. Unregistered data sources can only be used with Matching rules, not Deduplication rules.
    • (Optional) In Auto Accept Threshold, specify a match score that must be met or exceeded in order for a match result to automatically be accepted when matching is run. If a matching rule has a match score that meets or exceeds this threshold, it is displayed as Accepted in the match results screen and other matching candidates are displayed as Rejected. If multiple matching rules meet or exceed the threshold, the match result with the highest match score is accepted. If the match scores are the same for multiple rules, then the rule with the highest priority (that is, the lowest Rule Order) is accepted.

      Note:

      You can specify an Auto Accept Threshold for Matching rules only. You cannot specify an Auto Accept Threshold for Deduplication rules.
    • (Optional) In Auto Exclude Threshold, specify a match score that must be equal to or below in order for a match result to automatically be excluded when matching is run. Match candidates where the matching rule has a score below this threshold are not displayed in the match results screen.
  3. On the Definition tab, click Edit, and then click the + sign to specify rule criteria:
    • Use the Property drop down to select the property for the match. You can select from node-level properties with indexed values only.
    • Use the Match Type and Match Option drop downs to select the operator (and option, as needed) for the match. The match type values depend on the data type of the property that you select, and the match option values depend on the match type, as follows:
      Property Data Type Operators and Match Options
      String, Memo, Numeric String, Sequence
      • Equals
      • Contains
      • Similar To: Performs a fuzzy search on the property value.

        In Match Options, optionally specify a Prefix Length that specifies the number of characters at the beginning of the string that must exactly match. The prefix length can be used to reduce the query results to only those that have the same characters at the beginning of the string.

      Integer, Float
      • Equals
      • Between

        In Match Options, specify an offset to indicate the upper and lower boundaries of the range of values to match on.

        For example, if you are matching with an offset of 5 on a target property that has value of 10, then any incoming source property value from 5 to 15 will be considered a match.

      • Between %

        In Match Options, specify an offset percentage that the property value must be within of the source node's property value.

        For example, if you are matching with an offset of percentage of 15% on a target property that has value of 5000, then any incoming source property value from 4250 to 5750 will be considered a match.

      • Greater Than
      • Less Than
      Date, Timestamp
      • Equals
      • Between

        In Match Options, specify an offset (in days) to indicate the upper and lower boundaries of the date range to match on.

        For example, if you are matching with an offset in days of 3 on a target property that has a date of March 10, then any incoming source property value from March 7th to March 13th is considered a match.

      • Before
      • After
      Node data type Equals

      Tip:

      This identifies match candidates with a Node data type property that has the same property value as the source node.
    • When you have finished entering the first matching criteria, click + to enter a second criteria, or click Save to save the definition for the matching rule.
    • To remove a rule criteria, in the Action column click the Action menu action menu icon, and then click Remove.
  4. Click Save to save the criteria that you entered.

Deleting a Matching Rule

  1. Inspect the node type that contains the matching rule that you want to delete.
  2. From the Rules tab, select the Matching sub-tab, and then click Edit.
  3. In the Action column click the Action menu action menu icon for the rule that you want to delete, and then click Delete.
  4. Click Yes to confirm that you want to delete the matching rule.

    Note:

    You cannot delete a matching rule that has any matching history. If the matching rule is no longer needed, you can disable it instead.