Database design prospects opening with application of modern semantic data models

Semantic models and technique of their use at DB scheme designing are wrongly underestimated by many persons. Frequently semantic scheme is only an illustration of DB scheme created directly in DBMS model. In this case, those numerous problems of designing, about which this article narrates, are overcome without their deep analysis and use of the system approach. The article covers these problems of DB scheme designing and ways of their decision (it is possible automatic) with use of the modern semantic models - «Object - Role» (OR-model) and «Entity - Relationship - Mapping» (ERM-model). All modern DB scheme design techniques either initially assume creation of the integrated DB scheme, or integrate earlier created subschemes. Both in that and in the other case, the unification of representation forms and names of application domain (AD) phenomena is carried out. It leads to the loss of semantics located in subschemes, and its repeated finding-out at AD experts during creation of external user schemes. At the first ERM-designing stage data subschemes are created in exact accordance with requirements of subsystem users and with preservation of their representations and terminology. But they are defined not separately, and all together make the uniform ERM-scheme. At the following stage it is suggested to not refuse these data subschemes, and along with primary data forms and integrity constraints automatically to generate their base forms (classes and mappings) and to set interrelations between elements of various sub-schemes by the operations and relations between classes and mappings. In further, similar definitions will allow solving automatically many problems of designing. Passing from user subschemes to the integrated DB scheme, the designer faces one more problem: what data should be stored in the DB and what data should be derived from stored data automatically. Halpin distinguishes even three statuses of data, namely, stored, derived and semiderived. In OR-technique of DB designing the definition of data status is an exclusive prerogative of a person. He makes a concrete choice and he defines data derivation rules, fixing the decision in the OR-scheme. By this moment, the ERM-scheme already contains all interrelations of subscheme elements, and there is a hope that this information will be in most cases enough for the automatic decision of a problem «stored - derived». Also, it is possible to automate process of the derivation rule definition. All necessary information for it also is set in the ERM-scheme. After obtaining of the integrated scheme, it is necessary to design subschemes for separate users and user groups. Usually, during this stage designers again address to experts with the question, what they would like to see in dialogue with DB system and in what form. These representations are also necessary for realizing in DBMS. The basic tools of developers in case of relational DBMS are views and triggers. In the case of ERM-modeling external subschemes actually coincide with analytical ones. During designing elements of these sub-schemes are associated with their base equivalents and supplied with references to stored structures and integrity constraints. This information in the ERM-scheme is enough to completely define external user schemes in DBMS language. In the relational case, the generation process of the above mentioned views and triggers can be automated. For derived and semiderived data types rules of their calculation from stored data types are defined in views. Founders and researchers of semantic data models and DB designing technique give also the much attention to the bringing of AD semantics, fixed in schemes, to experts and users of the future system. There are features promoted the best understanding of the semantic scheme: - similar to human perception language of the scheme; - ability to present elements of the scheme as statements of a natural language (verbalization); - presentation of simple and clear data examples, satisfying and contradicting to the scheme (fact population). Use of the semantic technique supported by CASE-tools in which ideas offered in this article are realized, will allow to lead DB designing to qualitatively new and higher level.

Keywords

Authors

References