Semantic Reference Systems

Ontologies as Reference Frames

An ontology specifies the conceptualization of a particular information community by providing formal definitions for the terms used in this community. To do so, ontologies need to introduce primitive terms for the definition of more complex ones. For instance, the term river could be specified by introducing the term spring and the relation has_origin between rivers and springs. One could also define attributes, such as flow_velocity and water_gauge, and introduce kinds of rivers, such as forest_river. In turn, we also have to specify the meaning of terms such as spring and also the attributes and relations used to specify the meaning of river. We continue with the specification process until only primitives are left, that is, terms, relation, and attributes for which no formal definition can be given within our ontology. To ensure consistency in the interpretation of these terms, and hence to achieve semantic interoperability, we need to ground these primitives in observable reality by introducing semantic datums.

Summing up, ontologies restrict the potential meanings of terms used in an information community. However, to restrict the possible interpretations exactly to the intended ones, a semantic datum is needed. Besides specifying a vocabulary, ontologies support logical reasoning, which is a prerequisite for information retrieval. For instance, if we define particular attributes such as flow_velocity for river, we can infer that each water body that has been classified as a river has a flow velocity. Semantics-aware search engines use reasoning to deliver more flexible results. If a user types in stream as a keyword, the search engine would also include kinds of streams, such as rivers and creeks, and also similar water bodies (such as reservoirs if the user is querying for lake) in the result set.

...