The Surrogate Data Type

This short chapter introduces the SURROGATE data type, which is useful when modeling time-varying objects. Surrogates are unique identifiers that can be compared for equality, but the values of which cannot be seen by the users. In this sense, a surrogate is “pure” identity and does not describe a property (i.e., it has no observable value). For this reason, a SURROGATE data type cannot be treated identically to how other data types are treated. For example, at tuple cannot be assigned a specific value for a SURROGATE attribute. Rather, the user must indicate that the value to be assigned must be a new value never used before, or, alternatively, must indicate some SURROGATE attribute of some tuple that the assigned value must be identical to. For example, if surrogates are used for the identification of employees and a new tuple with information about an existing employee is to be entered then it is specified that the surrogate of the new tuple is to be identical to that of the existing tuple(s) for the particular employee. Beyond the semantics just mentioned, it is the responsibility of the user to give meaning to surrogates. For example, the user may, or may not, use attributes of SURROGATE type as keys. The SURROGATE data type is treated like any other data type when creating and altering relation schemas. An example follows.