Re: some information about anchor modeling

Hi Derek,
Thanks for your support. Your text contains a number of important questions related to db theory. I'll comment the three subjects that you defined here as "Surrogates”, “Plagiarism” and “Highly Normalized Tables". Today I'll write just about Surrogates. In the next few days I will comment the rest. First, I would like to clarify my opinion on Codd's contribution to the theory of database. I think that the RM is very important and it is well done. In my opinion, RM is a mathematical theory, and therefore, I think the theory of the database is a mathematical theory. I mean Codd’s greatest contribution is that he has devised the theory of database as a mathematical theory. So I agree with your opinion about importance of RM. However, I think that there is a certain group of people that unrealistically exaggerate Codd’s contribution to the theory of databases. I've always argued that the theory that explains the story about "relation-predicate-proposition" is Gottlob Frege theory and this mathematical theory is fundamental.

I agree with you that there is a poor use of the so-called IDs. But in this thread, I am writing about the surrogates used by Codd in his paper RM / T. Anchor Modeling and Object Oriented Approach also use the same surrogates. I also think that the RM / T is a bad paper, with many mistakes.

In this thread, I presented five examples that show fundamental weaknesses of the surrogates. I also explained, that would be a chaos, try to merge some two business applications, if their (different) surrogates denote the same entities. I write about these examples, because the first time in one place, they are clearly presented. There was some doubt earlier, some mistakes were observed. But problems with surrogates were never clearly and fully understood. The best-known experts in OO db, for years have tried unsuccessfully to resolve their problems. Now, my solution makes it possible to completely solve OO db. My solution also solves important problems in RM. For example my solution completely solves what Codd unsuccessfully tried to solve with RM / T.

But more important than these specific examples is the theory behind all of this. One of the important theoretical questions is identification, surrogates are about identification. This is about the identification of abstract objects. Some of my abstract objects have more than one identifier. Identifiers of the complex abstract objects exist only in the databases, similar to surrogates, but unlike surrogates, my identifiers are doing well. All objects that are stored in the human's memory or in the db's memory are abstract objects. In my solution, objects and attributes have real identifiers, while relationships and states have only identifiers in the database. Identifiers that are only in the database are linked to the identifiers that are in the real world. Note that in my solution, the attributes are treated as identifiers. You also can use my thread: Does the phrase "Russell's paradox" should be replaced with another phrase? Here I write about the identification of individuals and their relationship to the plurality. See also my post in this thread since 28.January, 2013, this is also related to identification. You can find my definition of abstract objects in my paper "Semantic Databases and Semantic Machines" section 1.1 at http://www.dbdesign11.com/

At the end, I want to say that my goal was to give a general procedure to allow proper db design that will allow the full solution to the problem of identification. Identification of objects in the real world and identification of the objects in a memory (remembrance) as well as vice versa.

Vladimir Odrljin Received on Sun Feb 10 2013 - 16:02:53 CET