Correlating heterogeneous schemas can range from specifying references across database boundaries to merging schemas in order to construct a global (integrated) view (Batini et al. 1986, Buneman et al. 1992)
Inter-database references based on data (e.g., WWW hypertext) links provide a simple form of correlating databases that does not require detecting or resolving conflicts. For example, one can specify that Maps in a mapping database are related to Sequences in a sequencing database by specifying an attribute for Maps representing the reference to Sequences. However, the data representing inter-database references (e.g., accession numbers) must be filled in each database, for all instances that have references to instances in other databases. Inter-database references entail ensuring the validity of these references when the databases change. Such inter-database referential integrity constraints are hard to maintain and are seldom enforced, which may lead to inconsistent references.
Integrating heterogeneous schemas usually involves some form of conflict detection or resolution (Batini et al. 1986) as well as schema and data restructuring. For example, conflict resolution includes establishing common (integrated) names, structures, and values. Schema and data restructuring are used for eliminating structural and data redundancies. Conflict detection and resolution can be prohibitively slow if they involve too many deciding factors, or unsatisfactory to some (possibly numerous) users if it involves only one or a few deciding factors.