Matching records locally

When you are creating or editing an import specification that is aligned with an i2 Analyze schema, you can tell the import wizard to use local matching to identify duplicate records in your incoming data, and with records on your chart.

About this task

Local matching is useful for identifying multiple records in the incoming data and on the chart. For example, a file that contains telephone call data is likely to refer to the same telephone more than once. In that case, you probably want the chart to contain a single telephone entity with links to other telephones, rather than lots of separate one-to-one relationships. Analyst's Notebook also tells you if information in your data import already exists on your target chart. If you do not need to search for duplicates in the Information Store, local matching gives you more control over the properties you are most interested in, as you define the matches during the import process. Also custom types cannot be matched with records in the Information Store. If system matching is unchecked on the Record Matching page, you can use local matching for system and custom types. Alternatively, you can check system matching to run at the same time as local matching. System types are matched against the rules defined by your system administator, while custom type matching is defined by you. Note that for custom types to match, they must have the same name and identifier.

Procedure

  1. Examine the Imported types box, which contains a list of the entities and links in your import design. The list includes both system and custom types. If you click the check box for system matching, any system types that are used in system matching rules are removed from the list.
  2. Select the type that you want to work with.
    The available options are different, depending on whether you choose an entity type or a link type.

    For entity types, you specify which property values must match in order for two incoming entities to be considered as duplicates. When the values of all the selected properties match, only one of the entities is imported. If a selected property has no value (but all of the other values match), you can decide whether to treat it as a duplicate.

    For link types, you have the additional option never to consider two links as duplicates, regardless of any property values. Furthermore, the list of properties implicitly includes the direction of a link. When two links have different directions, they are never treated as duplicates.

  3. Decide which property values determine what constitutes a match in your data.
    For example, it might be reasonable in some cases to match Person entities according to the combination of their given names and surnames. For data from a different source, you might use a passport number or social security number instead.
  4. Click the check boxes in the Identity column to specify the properties to use for matching.
    Imported types without any selected properties have a yellow triangle next to their name. If properties are selected, the triangle changes to the icon for that type. This is the case for both system and custom types.
  5. To prevent the import of less complete records, click the check box for Do not import a new record if any of the selected properties have blank values. This means that if any of the properties you select for matching have blank values, the import wizard ignores those records. This option is unchecked by default.

Results

When you run the import, a progress window is displayed. If matches are found with records on the chart, the import is paused. Another window is displayed that shows you the number of matches between your imported data and charted records. This information is to help you understand what is happening to your records during the import process. Duplicate records in the incoming data are not imported, if you chose to match on a specific property. For example, if your data contains 1000 records for the same main telephone number and you chose to match 'Main number', only 1 record for that telephone number is placed on the chart. If your incoming records contain more properties than the matched record on the chart, that record is enriched so the data is not lost.

What to do next

Click Continue to proceed with the import. When the import completes, a message is displayed to show timing details and the numbers of records that were created or replaced. The Analyst's Notebook records bar gives more information about the types of records now on your chart. Alternatively, you can click Cancel to end the import and return to the chart.