23.1. Extensional and intensional parts of the database

If you look closely at what is actually stored in the database, you will notice the presence of three different types of information. First, it is information characterizing the structure of user data (description of the structural part of the database schema). Such information in the case of a relational database is stored in system relations-directories and contains mainly the names of the basic relations and the names and data types of their attributes. Secondly, these are actually sets of user data tuples stored in user-defined relationships. Finally, thirdly, these are the rules defining the integrity constraints of the database, database triggers and represented (virtual) relationships. In relational systems, the rules are again stored in system tables-directories, although flat tables are far from ideal for this purpose.

Information of the first and second types in the aggregate clearly describes the objects (entities) of the real world, modeled in the database. In other words, these are explicit facts provided by users for storage in the database. This part of the database is called extensional.

Information of the third type serves to guide the DBMS when performing various kinds of operations defined by users. Integrity constraints can block the execution of database update operations, triggers cause automatic execution of specified actions when specified conditions occur, view definitions cause explicit or indirect materialization of presented tables when they are used. This part of the database is called the intensional; it contains not immediate facts, but information characterizing the semantics of the subject domain.

As can be seen, in relational databases, the extensional part is most important, and the intensional part plays a mainly supporting role. In rule-based database systems, these two parts are at least equal in rights.