Semantics in Data and Knowledge Bases

Semantics is the study of meaning, i.e. how meaning is constructed, interpreted, clarified, obscured, illustrated, simplified, negotiated, contradicted and paraphrased [Wan87]. It has been treated differently in the scientific community, e.g., in the area of knowledge bases and by database users.

The scientific community prefers the treatment of ‘always valid’ semantics based on the mathematical logic. A constraint is valid if this is the case in any correct database.

Database modellers often use a ‘strong’ semantics for several classes of constraints. Cardinality constraints are based on the requirement that databases exist for both cases, for the minimal and for the maximal case.

Database mining is based on a ‘may be valid’ semantics. A constraint is considered to be a candidate for a valid formula.

Users usually use a weak ‘in most cases valid’ semantics. They consider a constraint to be valid if this is the usual case.

Different groups of users use an ‘epistemic’ semantics. For each of the group its set of constraints is valid in their data. Different sets of constraints can even contradict.

Semantics is currently one of the most overused notions in modern computer science literature. Its understanding spans from synonyms for structuring or synonyms for structuring on the basis of words to precise defined semantics. This partial misuse results in a mismatch of languages, in neglecting formal foundations, and in brute-force definitions of the meaning of syntactic constructions.