Null values are used to represent uncertain data values in a database instance.
Since the beginning of the relational data model, null values have been investigated, with the intention of capturing and representing data values that are uncertain. Depending on the intuitions and cases of uncertainty, different kinds of null values have been proposed, e.g., they may represent information that is withheld, inapplicable, missing, unknown, etc. Thus, in principle, it could be possible to find in a hypothetical database diverse classes of null values, and also several null values of the same class. However, in commercial relational DBMSs and in the SQL Standard, only a single constant, NULL, is used to represent the missing values.
Many semantic problems appear when null values are integrated with the rest of the relational data model, which essentially follows the semantics of predicate logic. Among them, (i) the interpretation of nulls values (for a particular...
- 1.Grahne G. The Problem of incomplete information in relational databases. LNCS, vol. 554. Secaucus: Springer; 1991.Google Scholar