Complex networks can be represented in a variety of ways. They can be decomposed into trees or sim-ple networks, and these simpler structures can then be represented using one of the techniques just described. Alternatively, they can be represented directly using indexes. Linked lists are not used by any DBMS product to represent complex networks directly. In practice, complex networks are nearly always decomposed into simpler structures, so we consider only those representations using decomposition.
A common approach to representing complex networks is to reduce them to simple networks and then to represent the simple networks with linked lists or indexes. Note, however, that a complex network involves a relationship between two records, whereas a simple network involves relationships among three records. Thus, in order to decompose a complex network into a simple one, we need to create a third record type.
The record that is created when a complex network is decomposed into a simple one is called an inter-section record. Consider the StudentClass complex network. An intersection record contains a unique key from a STUDENT record and a unique key from a corresponding CLASS record. It will contain no oth-er application data, although it might contain link fields. The general structure of this relationship is shown in Figure H-29. Assuming that the record names are unique (such as S1, S2, and C1), an instance of the STUDENT-CLASS relationship is illustrated in Figure H-30.