Re: The original version
Date: Fri, 7 Jan 2011 14:13:53 -0800 (PST)
Message-ID: <8e74aacf-7710-4ee5-9ec5-8a07b85e19e5_at_s5g2000yqm.googlegroups.com>
(i)
As far as I can see, Anchor Modeling cannot maintain the history of a
database regarding any kind of deleting of data. For example, there is
an option “logical delete” in many old software, that can maintain the
history of a database. Note that there are only three operands –
delete, insert and update - which are related to changes of data.
(ii)
The question is how to construct basic structures in Anchor Modeling?
How to construct "Historized Attributes"?
First one has to construct an entity, which entity may not be actually
composed of two other entities. The entity must comply with the
conditions that are consistent with "normal forms". None of this is
solved in Anchor Modeling, but it is in my paper.
Secondly, if an entity is constructed, then we can apply anchors and
get “Historized Attributes”. So, I have the entity E, and from it I
get "Historized Attributes" HA1,...,HAn. An important thing is how to
get the entity E from HA1,...,HAn. I solved this using relationships,
i.e., using only the ER model (see Example 6 in my paper). Anchor
Modeling cannot solve this problem using the ER model.(see my message
from 24 December in this thread, where I explained that Anchor Model
cannot solve historized relationships).
Also it is necessary to explain how to decompose a relation into
binary relations (also solved in my paper - see sections 4.2.6 –
4.2.9, 6.4 and 6.5). Only after these constructions of binary
concepts and binary relations is it possible to do a mapping between
them.
6NF cannot do this as the authors explained in first version of Anchor
Modeling. Anyone who is interested, can find a critique in this thread
related to this argumentation of the authors of Anchor Modeling.
Now, in the second version of Anchor Modeling 6NF is omitted from the
title of the paper. In this new version, I can not find reference
[19], which was about NFs.
In the second version of their paper, the authors change the
definition of “Historized Attribute”.
They introduce by definition all what I did regarding binary
structures. And I really don’t have time for this kind of “theory”.
Vladimir Odrljin Received on Fri Jan 07 2011 - 23:13:53 CET