[Data-modeling] software versions/releases in the computer domain?

Tue Apr 29 22:12:48 UTC 2008

Brendan - Sorry for the confusion. I was addressing an underlying theme I've
encountered in Freebase data models rather than your particular brand of it.
Many times these modeling discussions begin because an existing data model
requested a single value for a dynamic property. Maybe these properties are
useful placeholders for quick and dirty models to get the ball rolling, but
they could possibly cause a lot of problems down the road. This is
especially the case for public models. It would be nice if we all gave
greater consideration to whether a property is dynamic or static during
model development. Maybe there could be a default "dynamic property"
property that could serve as a better placeholder?  I usually see these
improperly modeled dynamic properties set as text or floating point values.

Tangent: it is also important to note when values are not absolute. For
example, if a value is an estimate then the schema should allow the user to
list various statistical summaries. Otherwise, we end up with the means
listed sometimes, max others, and no way to keep track of variation - which
is often just as important.

While it is always useful to research how other people describe topics using
other methods and models, Freebase seems to have a nascent ontological
physics that is all its own (in my limited experience anyway). Some
translation between Freebase schema and other data modeling structures will
likely always be required.

Your need for a simple but useful hierarchy to document other's coding
systems reminds me of the discussion about modeling Library of Congress and
Dewey codes!

-Ed

On Tue, Apr 29, 2008 at 4:01 PM, brendan <brendan at metaweb.com> wrote:

> Ed,
>
> I think I understand what you're saying (but would like you to clarify). I
> think what you're saying is there are probably existing examples of modeling
> this kind of data outside of freebase ("everything that came before") and I
> should survey these approaches and take a more rigorous approach in
> proposing schema (e.g. use references, compare and contrast existing
> approaches, etc).  I can't argue with that. It's prudent. I think you are
> also saying I should look at the existing schema and data in freebase.  I
> have (though I didn't document it) and I found it pretty much absent of
> structure/data around software releases/versions (except for a small amount
> of operating system data)
>
> With regard to "Major Release" I must admit that I find myself trying to
> reconcile to what exists in the wikipedia and that can be constraining.  In
> this example, the wikipedia has an article for:
>
> Internet Explorer 4
> Internet Explorer 5
> Internet Explorer 6
> Internet Explorer 7
>
> These topics do not describe discrete software releases/versions, rather,
> they describe a generally accepted grouping of releases that, apparently,
> wikipedians think might be important.  I think that's because these
> groupings have real utility in the world: websites don't usually claim
> compatibility to a particular release version but rather the range (e.g.
> Google Gmail browser support doc
> http://mail.google.com/support/bin/answer.py?hl=en&answer=6557)<http://mail.google.com/support/bin/answer.py?hl=en&answer=6557%29>
>
> That said, I like your flat approach. As I mention above, I find myself
> overly influenced by the way wikipedians slice up data.  I guess the anxiety
> comes down to: if I were to implement your flat approach what the heck do I
> call that list of topics above (the IE releases)?  They're not software
> versions, as I stated, they're something more abstract. Should I just leave
> them alone?  But then other users will get confused at their presence.
> Should I delete them?  What I'm saying is this is a case were Wikipedia has
> a sort of implicit schema and if we don't find it satisfactory we need some
> way to reconcile with it.
>
> Brendan
>
> On Apr 29, 2008, at 12:10 PM, Ed Laurent wrote:
>
> I echo Tim only because he beat me to it!
>
> Brendan, I think you've bumped into another instance of Freebase focusing
> on "current" information while ignoring everything that came before and not
> documenting the currentness (it's a word, look it up<http://dictionary.reference.com/browse/currentness>!)
> of the data. While this certainly keeps the schema simple and approachable
> to a wide audience, I'm not sold on it being the best default approach to
> modeling over the long term.
>
> I also think you are on a good track for modeling software releases within
> a simple hierarchical framework. However, I would take a more systematic
> approach to the releases (even if that requires disambiguation between the
> date of release and version number). You could then treat every release the
> same but co-type certain releases as a "major release" with properties that
> define why it is a major release. However, distinguishing what constitutes a
> "major release" could lead to a complex model that is never used unless it
> is very generic or software specific (e.g., Firefox major release).
>
> -Ed
>
>
> On Tue, Apr 29, 2008 at 2:58 PM, Tim Kientzle <tim at metaweb.com> wrote:
>
> > I like the idea of keeping the heirarchy here shallow and with a fixed
> > number of levels:
> >
> > "Software Product"  -> "Major Software Release" -> "Release"
> >
> > Detailed modeling of major/minor/revision/bugfix is troublesome
> > because there is so little consistency across products.  Open source
> > packaging systems have major challenges just trying to consistently
> > order versions (consider a product that switches to year-based version
> > numbering then back again, so that 2.0 precedes 2006.0 precedes 5.4,
> > not to mention ambiguous numbering such as "Solaris 7" which was
> > really 2.7 except that the marketers decided to drop the '2'.)
> >
> > In fact, I'd be wary of using the words "Major" and "Minor" because
> > they do carry a lot of baggage.
> >
> >
> > On Apr 29, 2008, at 11:28 AM, brendan wrote:
> >
> > > hopefully your in-brain word wrap filter is in working order :(
> > >
> > > let me try that again:
> > >
> > >
> > > The "Major Software Release" type could have a property "Releases"
> > > with an expected type "Software Release", so every actual versioned
> > > release could have a topic
> > >
> > > e.g.
> > >                                                              ---> 1.0
> > >                                     ---> Firefox 1 ---> 1.0.1
> > >                                                             ---> ...
> > >                                                             ---> 1.5
> > > Mozilla Firefox --->
> > >                                    ---> Firefox 2 ...
> > >                                    ---> Firefox 3 ...
> > >
> > > _______________________________________________
> > > Data-modeling mailing list
> > > Data-modeling at freebase.com
> > > http://lists.freebase.com/mailman/listinfo/data-modeling
> >
> > _______________________________________________
> > Data-modeling mailing list
> > Data-modeling at freebase.com
> > http://lists.freebase.com/mailman/listinfo/data-modeling
> >
>
> _______________________________________________
> Data-modeling mailing list
> Data-modeling at freebase.com
> http://lists.freebase.com/mailman/listinfo/data-modeling
>
>
>
> _______________________________________________
> Data-modeling mailing list
> Data-modeling at freebase.com
> http://lists.freebase.com/mailman/listinfo/data-modeling
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.freebase.com/pipermail/data-modeling/attachments/20080429/7f775fc2/attachment-0001.htm 