[1] Semantic translation takes advantage of semantics that associate meaning with individual data elements in one dictionary to create an equivalent meaning in a second system.
This is frequently required by intelligent agents that wish to perform searches on remote computer systems that use different data models to store their data elements.
Semantic translation requires that data elements in the source and destination systems have "semantic mappings" to a central registry or registries of data elements.
There are three types of Semantic equivalence: Semantic translation is very difficult if the terms in a particular data model do not have direct one-to-one mappings to data elements in a foreign data model.
This problem can be alleviated by centralized metadata registries that use the ISO-11179 standards such as the National Information Exchange Model (NIEM).