On the Design, Use, and Integration of Data Models.
Abstract
In spite of the large number of data models that have been proposed, no consensus exists about what concepts are really important for data modelling. Presentations of the various models cite the advantages of their approach, and discuss what the model best can represent. The advantages of different approaches when they include ease of representation, ease of implementation, naturalness of expression, power of query language, non-procedurality of query language, expression of structural constraints, efficiency of implementation, formal foundations, support of existing implemented models,... etc? Another problem encountered when we attempt to compare models is that not all models are defined at the same level. For example, hierarchical and network models include numerous implementation details. The relation model does not include any inter-relation constraints. Although some real-world constraints are captured by functional, multi-valued, and other dependencies, no systems represent these constraints except the functional dependencies that result form the primary keys of relations. The entity-relationship model is claimed to be a real-world model, rather than an information structure model, as are semantic and binary relationship models - does that imply these models cannot or should not be used for database implementation? This thesis, is concerned with data models at the information structure level, rather than at the real-world semantic level. However, we want the information structure data model that we use to be able to capture many of the semantic concepts of the real-world that it models, as well as to support a query language which is close to the way users perceive the real-world to be organized (to a certain extent).
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 1980
- Accession Number
- ADA091384
Entities
People
- Ramez Aziz El-masri
Organizations
- Stanford University