Data models

TL;DR

This article surveys database data models, classifying them into object-based, record-based, and physical categories while focusing on logical models. It describes the entity-relationship, object-oriented, relational, network, hierarchical, and object-relational approaches and notes a trend toward combining relational and object-oriented features.

Key findings

• Data models are grouped into three broad categories: object-based logical, record-based logical, and physical; the article sets physical models aside because they concern implementation details.
• Object-based models include the entity-relationship model, with entity sets, relationship sets, attributes, primary keys, and extensions such as specialization, generalization, and aggregation, and the object-oriented model, with encapsulation, instance variables, messages, methods, classes, and inheritance.
• Record-based models structure data as fixed-format records; the relational model is described as the leading commercial model, grounded in mathematical relations and tables, with relational calculus and algebra as formal query foundations and SQL-92 as the established standard.
• The network and hierarchical models are tied more closely to physical design and are declining in importance, while object-relational hybrids extend the relational model with complex types, object identity, and inheritance; SQL-3 was under development with object-oriented features.

Methods

• Conceptual survey and taxonomy rather than an empirical study; it organizes and compares data models by their descriptive primitives and semantic expressiveness.
• No dataset or sample is used; the discussion draws on prior database literature and texts, including work by Chen, Codd, Kim, Maier, and Silberschatz et al.
• Main analysis is descriptive: it explains modeling concepts, query languages, standards, and convergence trends across model families.

Limitations

• The paper states no limitations. Evident from the text: it offers no empirical evaluation, comparative metrics, or systematic review protocol.
• Evident from the text: coverage is brief and high-level, and physical data models are explicitly excluded from the discussion.
• Evident from the text: the object-relational discussion is intentionally short, and the 1996 publication date makes its standards and market observations historical rather than current.