RE: REA (was Tags and semantics ( was Dotted Names) )

[Bob Haugen <linkage@interaccess.com>]

John McClure:
>Bob Haugen says:
>> Many people within the ebXML BP group wanted to get REA embedded
>> in the first version of the ebXML runtime specs, but we ran out of time,
>> bandwidth, etc. and I think also wanted to keep the required
>> implementation specs more modest.

>I can understand the position as far as the business process model is
>concerned, but the REA data model I would think would be a major topic
>within the CC group. Doesn't seem to be, otherwise you'd find a taxonomy
>whose fundamental classes include AT LEAST "Resource", "Event", and "Agent".
>Instead, I see "Party". But maybe I've missed it -- it wouldnt surprise me
>given the avalanche of ebXML's documentation!

No, as far as I know, nobody in the CC group has considered
the REA semantic model seriously, although some of us have tried
to promote it.  I suspect people have had enough to do with harvesting
data components from various sources. The taxonomies you mention
would help.  The context classifications included "Product", by which
the CC group meant both product and economic service but not cash.
Economic Events were not specified in the CC context classifications.
"Party" = "Economic Agent", as you mentioned, and I agree that the
word Agent is too overloaded to be useful in a technical spec.

I do think the information model, process model and business
semantics need to come together in some next-stage activities,
and will do all I can to see that this happens.

>
> >REA is embedded within the DCN's list of 15 resource-types. I have the
> >suspicion that the same cannot be said for the 600+ DTDs in
> xCBL, although I
> >haven't looked into it. Should I?
>
> That news about DCN is very interesting; can you point out to me
> where to look in the DCN references?
>

[interesting details about DCN snipped, but I will study them, thanks.]

>The point of this litany is to demonstrate that REA has a data-model as much
>as anything else. 

Yes, REA started out as a data model, back in Entity-Relationship days.

>The design of the DCN makes a reasonable stab at it -- I
>hadn't read about REA until just 2 weeks ago, but I think that the DCN
>arrives at the same place as would be if a "REA Data-model" existed. 

I always think of REA as a semantic model abstracted from 
operational business systems, so it would not surprise me that
parallel efforts would end up in similar places. 

What I like about REA (among other things) is that Bill McCarthy
and colleagues have been refining it for 20 years, and they have
been determined to keep it as minimal as possible, so it's well-
founded and clean.

>While
>it's certainly ok to defer the REA Process-model for the time being, I think
>it is ill-advised to  defer its data-model, being as fundamental as it is to
>system development.

I agree on its importance.  Got any action proposals?

>Now, could you answer my question about the 'fit' between REA and xCBL?

No, sorry, I haven't studied xCBL enough.

Regards,
Bob Haugen