Classes, Entities and Tables are much too coarse to be useful in Business Information Modelling, where Concept Maps and Business Rules are preferable. Actually a Table is a designed artefact, which contains generalisations - and not many business rules. These facts I have been preaching for many years and a place to start is here.

Moving the “granularity” one level down, to the concept / attribute level means you can do things, you could not do before. Here I will mention two emerging kinds of solutions:

1) New DBMS’s, which work on the atomic level. They come in a variety of flavors: Columnar databases, Graph databases (incl. RDF), Associative databases and more. One can also argue that the Data Vault / Anchor Modelling approaches are in this ballgame, too. The newest kid (DBMS) on the block is the “Algebraic Database” - a product called A2DB from Algebraix Data. Robin Bloor has written an excellent white paper about the paradigm and the technology.

2) The Semantic Web approach is certainly gaining acceptance these years. RDF actually works on the atomic level (subject-predicate-object). The most visible (and largest) “killer app” for RDF, OWL and SPARQL is the Linked Open Data movement. More than 200 public databases can be linked (transparently) thanks to the shared standards and shared usage recommendations. Actually, Linked Data is also an attractive technology for enterprise level integration of data. And it is being used for that, today. A new approach to Enterprise Information Management, indeed.

Both of these issues are very important, and I will come back to them on the regular pages of this site.

Tags: Columnar databases, Associative databases, Linked Data, Database technology