Atomicity and Semantic Normalization

Halpin, Terry; Carver, Andy

doi:10.4018/jismd.2010040102

Cited by 3 publications

(2 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It is arguable that the approach to the construction of a DB scheme using decomposition is unstable in a sense with respect to the initial set of attributes and relationships. Alternative synthesis algorithms for normalized relational schemes according to Bernstein fail if the semantics of an object domain presumes the presence of multivalued dependencies or connection dependencies.Normalization increasingly seems to be something imperfect and requiring a cardinal revision [3,4]. At the same time, the relational model evolves by including approaches based on models of universal and binary relations, the 1008…”

mentioning

confidence: 99%

See 1 more Smart Citation

Properties of a relational framework synthesized over a set of semantically atomic predicates

Панченко

Pysanko

2009

Cybern Syst Anal

View full text Add to dashboard Cite

Properties of a relational framework synthesized over a set of semantically atomic predicates are analyzed. A relational framework is shown to be stable with respect to a modification in the basic predicate set. The stability, uniqueness, and completeness allow one to consider a relational framework as a universal "carrier" for object domains with arbitrary semantics.Keywords: relational framework, semantically atomic predicate, scheme of a relational database, cortege, domain.A scheme of a relational database (DB) is traditionally understood to be some fixed collection of relational schemes R j , i.e., named sets of attributes and keys [1]. As a rule, for the construction of such a scheme, a collection of attributes x i and sets of values, i.e., domains D x i ( ), that are uniquely associated with these attributes are introduced [2]. In this case, collections of the attributes themselves are associated with objects or entities, and collections of values of attributes are associated with instances of objects or entities. It is the first step to the mapping of the semantics of an object domain into the scheme of a DB. Note that the set of x i and the collection of sets D x i ( )are common for schemes R j in the sense that a separate attribute can belong to several schemes. Finally, an instance of each relational scheme R j is represented in the form of a collection of corteges K p , i.e., ordered sequences of values of attributes x i of the jth scheme R j , and we haveHeterogeneous interrelations revealed between sets of attributes by the analysis of semantics of object domains, play the decisive role during the normalization of the scheme of a DB. It may be said that, in the approach "attributes-domains," an amorphous "field" is first specified over which the revealed semantic relationships make it possible to form (after a number of transformations that is, as a rule, rather large for nontrivial examples) a normalized structure of a DB. This "formation of a normalized structure" is one of the most labor-consuming stages of developing the logic scheme of any DB.Despite the existence of advanced techniques of decomposition of DB schemes, this approach to design has a cardinal drawback [2] , namely, the absence of flexibility. It is a change in the collection of attributes and/or interrelations between them (for example, after an extension of the object domain of a DB, which mostly happens in practice) that causes the necessity of repeated formation of the normalized scheme of a DB. In such cases, the logical scheme of a DB is difficultly modified or cannot be modified at all (since there is no strict continuity between old and new schemes of the DB). To represent new (additional) objects, entities, and relationships in a DB, it is required to form an extended collection of objects and relationships and then again perform the normalization procedure, i.e., to practically again solve the decomposition problem. In this case, a correct procedure of normalization of the scheme of a DB must provide data security, i.e., ...

show abstract

mentioning

confidence: 99%

“…Normalization increasingly seems to be something imperfect and requiring a cardinal revision [3,4]. At the same time, the relational model evolves by including approaches based on models of universal and binary relations, the "entity-relationship" model, and object-oriented and semantic models [1,2].…”

mentioning

confidence: 99%