Assumed semantic types for labels

You can use the text in the label of an item for matching in addition to any values in attributes or data records. Depending on the item type, a semantic type is assumed for the label value. It is not possible to alter which semantic type is used.

The table shows how the labels of entities are treated for matching purposes. It shows, for example, that labels on Location entity types are treated as the Location Name and not the address.

Semantic Type Assumed Semantic Type for Label
Person Person Details
Organization Organization Name
Phone Phone Number
Motor Vehicle Motor Vehicle Details
Location Location Name
Event Event Title
Bank Account Bank Account Details
Bank Card Bank Card Details
Website Web Address
Mailing Address Full Address
Drivers license Drivers license Number
Bank Bank Name
Vehicle Registration Vehicle license Plate Number

Some examples of how labels are used:

Person entity types

In Smart Matching, single-word names get a low score; they are not displayed as a Matched Set unless they have further information with matching behavior. For example, if a person with a single-word name has a date of birth, or has other properties that match, they are displayed as a Matched Set in Find Matching Entities.

Motor Vehicle entity types

In Smart Matching, labels on Motor Vehicle entity types are treated as Motor Vehicle Details and a match score might be generated if a label contains a license Plate Number. Other property semantic types might be recognized from a Motor Vehicle label but no match is found unless a license Plate Number is detected in the label, an attribute or a property.

Location entity types

In Smart Matching, places and businesses (bars, shops, and so on), are assigned the Location semantic type. The labels are treated as location names and not addresses. To match addresses the entities need to have further address information with matching behavior, for example, the first line of an address or a zip code. This information could be supplied as attributes or in data records.