Hi,

Hopefully I'm posting this in the correct place. I've searched the forum / browsed the sticky topics prior to creating this thread.

I've been assigned the task of developing a system based on the associative model. There doesn't seem to be much information available on this topic. I've already looked at http://www.lazysoft.com/docs/other_docs/AMD.pdf and Lazysoft's Sentences program, which seem to be the best material I can find.

So, the ideas all make sense, but now I'm coming to implementation... I have the following points to consider:

1. How will the DB manage unique IDs accross the Item / Link tables? It's not sufficient to simple declare that each table should have a unique ID. The surrogate IDs of records should be unique accross both tables. I can think of ways to do this programmatically, but nothing on a DBMS level... I'll be using MSSQL.

2. What form will queries take? I'm well versed in SQL when it comes to relational DBs, but this is new territory for me... If the Link table is just a table of linked IDs, I cannot explicitly declare literal values... Furthermore, should I be trying to declare literal values? E.g. "Select ((((“Mary” customer of Store) visited on Date) bought Product)
join (Product category “Dairy”))" - example taken from http://www.lazysoft.com/docs/other_docs/AMD.pdf @ page 186. The SQL itself is simple in terms of set theory, but I can't see how to get a query like that in / out of the DB. I can only imagine there is some intermediary layer where the link table is displayed using names not IDs, or that the SQL has some clause to join Link.SourceID, Link.VerbID or Link.TargetID to Item.ID where Item.Name = 'Some Value'. Hmm...

Sorry for the vague / confused / confusing nature of this post... As previously mentioned, this is all a bit new to me. ANY examples of associative DB implementation, additional reading or example SQL would be beneficial.

Thanks in advance.

Why do I get the feeling this is outside of everyone's comfort zone (including mine!)?

I think the same.

What was the process that led to the business decision that it would be ideal to develop a database in a format the developer has no experience of? You sound comfortable with the relational model and SQL. Or is this academic?

pootle flump ur codings are working excelent.

Simon Williams came and tried to promote his Associate Model here and here on dbforums about 6 years ago. Only one of the regular members was impressed, and he proved himself to be certifiably bonkers in later posts and blogs. So it's not an idea that has much traction round here!

Academic-industrial motivation...

Academic: I'm a developer coming to the end of a work placement which forms part of the degree course I'm completing. This is supposed to be a bit of a challenge, which is fine...

Industrial: The project in hand lends itself quite well to this sort of thing. It's essentially a lightweight configuration management system to model various items / entities within the company and how they interact / depend on each other (all types of associations for a given item). As it happens, the associative model is apparently very good at storing these kind of links, and lends itself well to trees and such, which will be useful for browsing the company infrastructure. If anyone has seen Sentences you'll see what I mean about the tree structure...

As previously stated, the concepts all click, and the approach is sensible (lack of direct experience aside). I'm just tripping up on the queries a bit, but that's because I have seen 0 actual examples to get me started, and I don't really want to try to invent best practices... I might have to hit the library / book shop methinks.

EDIT: Just seen the second response... I'll take a look into those links. Thanks! So this is officially the dark side approach...

Yes, but to be fair to it, it would be interesting to hear from the experiences of someone like yourself trying to actually use it. The LazySoft web site has plenty of references to endorsements from "industry analysts" like Bloor and Butler, but I don't have much respect for such pundits - they just seem to like to jump on new bandwagons and don't necessarily understand the subjects in any great depth. I'm sure they are/were all in favour of "object databases", "post-relational databases" etc. too!

I seem to be feeling particularly cynical today?

Hmm, I didn't originally expect this to be such a pioneering exercise in implementation... I did wonder why there was such limited material available. I could potentially construct an argument as to why, as sole developer, I don't think it's feasible to continue with the associative model and revert back to the relational model I documented early on... But at the same time, I think I'll at least have a good shot at trying.

Thanks again for your links - they've been insightful so far, and it's interesting to see someone (you!) having a good, constructive dig at the whole approach. Well, I guess this all just makes it even more academic... xD

I'll aim to keep this updated if / when I make progress then.

Cheers.

Tom, I got your message but I'm unable to reply by email as the forum says you don't accept them. I'd be happy to help - email me at simon.williams@lazysoft.net.

...followed by this classically ridiculous thread here:
http://www.dbforums.com/thrown-threa...hitecture.html
Long story short, you won't find much respect for the Associative Model or any other impractical implementations on dbforums. I suggest you contact Certus, or one of the other crackpots directly. Hope you have a lot of time on your hands and enjoy diagrams of the five platonic solids.

ROFL, Blindman! (And I've never said ROFL before). I'd forgotten just how funny that thread was - thanks for reminding me. I must make a note to read it again in about 2 years time...

too bad we mere mortals cannot enjoy a thrown thread

Actually, it's too funny. My inability to stop myself from guffawing and crying with laughter while re-reading that thread at work has caused me extreme embarassment. Must remember to read it only when at home in future.

Yes, isn't it? I wonder if it can be "unthrown"?

Moved, and locked.
Associative Model of Data Architecture
Enjoy the surreality, and enter the Phantom Tollbooth of Database Design.

awesome, thanks

i, too, enjoyed a brief dip in the pool of insanity...
if he's paying attention (that's a big if) he will notice new hits to his web site from dbforums, and who knows, maybe he'll come back and let us know what else he's managed to accomplish since we saw him last...