I hesitate to get involved in this, but…
RDBMS refers to a particular style of representing and using ‘relations’, namely as ‘tables’.
There are a lot of tree-structured and graph-structured databases out there, some born of direct use of XML files as information repositories (“databases” in
a general sense). While one can say that an arc of a graph or a ‘containment’ in XML or a ‘branch’ of a tree structure also represents a ‘relation’, it is a more general meaning of the term than David had in mind. In a similar way, RDF represents relations
by triples, but most triple stores don’t sort by the ‘verbs’ that would make them ‘tables’, but rather by the ‘nouns’ that play the roles in the verb triples. (Of course, RDBMS use ‘indexing’ to achieve the same effect.)
So this is really all about the definition you assign to the term ‘relation’. By comparison, the term ‘relational database’ is pretty well defined by 35 years
of practice. It is not just SQL, but there is value to having a standard. (The problem there is that the most recent version of ISO 9075 has 41 other information structures and ancillary programming and data formation languages, in the unceasing effort to
be all things to all men, lest the vendors lose some fraction of their market.)
BTW, in the ontology community, there are two or three interesting kinds of ‘mapping’ as well. One of them is about relationships between formal ontologies,
one of them is about mapping data schemas to formal ontologies, not to be confused with ‘lifting’ data schemas by rote to a language like OWL, with no clear sense of the conceptual intent (“semantics”). Finally, as Bill Burkett pointed out, there is the idea
of “data mapping”, which is about converting “instance data” from one representation to another, where some of those representations might be in “knowledge representation languages”. Perhaps the most interesting version of that is the idea of “semantic mapping”
-- source data to KB representation, then KB representation to target representation, all based on the idea of maintaining the intent of the data, using a formal representation of that intent in a KR language.
From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx [mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx]
On Behalf Of Kingsley Idehen
Sent: Tuesday, February 17, 2015 10:46 AM
Subject: Re: [ontolog-forum] master data vs. ontologies
On 2/17/15 9:20 AM, David Eddy wrote:
The instances are represented by the rows in relational tables.
I'm uncomfortable with the repeated focus on relational databases.
What about non-relational structures?
What about those single table, 2000 column RDBMS applications?
Does any such thing as a non-relational database management system actually exist, bearing in mind data is represented using relations?
Circa., 2015, can we seriously allow SQL-RDBMS marketing-driven dogma turn the very notion of relations and database management on its head?
Relations represented as sets of Records in a Table isn't the only way to represent relations (that's SQL RDBMS dogma). You can also have them represented as Relational Predicate/Property Graphs (sets grouped by sentence predicate).
http://kidehen.blogspot.com/2015/01/loosely-coupling-database-document.html -- Loosely Coupling Database Management Systems & Database Documents
http://kidehen.blogspot.com/2014/12/deceptively-simple-conceptual-data.html -- Deceptively Simple Conceptual Data Virtualization
http://kidehen.blogspot.com/2014/08/linked-local-data-lld-and-linked-open.html -- Linked Local Data vs Linked Open Data
http://kidehen.blogspot.com/2015/01/from-open-database-connectivity-to.html -- From Open Database Connectivity to Open Data Connectivity .
Founder & CEO
Company Web: http://www.openlinksw.com
Personal Weblog 1: http://kidehen.blogspot.com
Personal Weblog 2: http://www.openlinksw.com/blog/~kidehen
Twitter Profile: https://twitter.com/kidehen
Google+ Profile: https://plus.google.com/+KingsleyIdehen/about
LinkedIn Profile: http://www.linkedin.com/in/kidehen
Personal WebID: http://kingsley.idehen.net/dataspace/person/kidehen#this