[moby] [MOBY-dev] Reserved word... perhaps unnecessary...

[Andrew D. Farmer]

Hi Mark and Martin-
I can't resist jumping in here from somewhere out of left field (and 
without much time over the next week to engage very deeply in the discussion), 
BUT:

it seems to me that the argument you are beginning to hash out here 
gets close to the issue that the semantic moby side of the world has 
with respect to moby services' MOBY-XML representation of data vs semantic
moby's preference to represent the data (not just the ontologies) as 
RDF (using RDF/XML or some other RDF syntax). 

The position below that you ascribe to Martin concerning the non-essential
nature of the objects would for the semantic web data representation approach
be instead better characterized as a "multi-faceted essence"; that is to say 
that an object can belong to many classes and each of these "types" 
associated with the object may (or may not) imply a
set of properties ("the stuff") to be expected of an instance of the
type OR (conversely) an object that has a certain set of properties may
be inferred to belong to a class defined in a certain way. The decision
about what is "essential" is I believe one determined by the user of the
information, and the best systems are those that allow the same "stuff"
to be used in many different purposes (since there is no single type
of user). I think it is true that MOBY objects can have multiple
typing, by reusing the namespace id in a structure with a different "class tag"
(apologies if I have the lingo a bit off, it's been a while for me); and
yet, I think your insistence that the ISA determines the object's essence
is somewhat indicative of the use of XML structure to have the data producer
determine intended usage rather than allowing the data consumer to determine
how it may be used. (If we wanted to make this discussion more concrete,
I have been fooling around with a little OWL ontology ("owl-let") for
properties representing mapping/sequence/feature type info; it is in the
same conceptual space as your "genericsequence/virtualsequence/etc." stuff,
and might be an instructive example of how the two approaches differ
in practice.)

My 2c, and not necessarily reflective of the opinions of others in the
s-moby sphere, but I do think that this is one of the critical areas to resolve
in order to be able to bring the two branches of the project back together.
Whether or not RDF is appropriate as a representation of all
pieces of a message is another question, but in terms of a data payload
representation format, I think we should really give it some 
consideration if and when we are able to move forward on the long promised
reunification of the two branches. 

RDF: it's not just for metadata!

Andrew Farmer

PS. I hope this message doesn't hit Martin's mailbox before he's had time for
tea... ;)

PPS I am not clear on the "rendering" issues you allude to at the end,
but we have been having some debates around the issue of transmitting
data in well-known text formats embedded within (or as some sort of
attachment to) the RDF. I would be interested in hearing about your 
experiences with such things...

On Thu, 3 Nov 2005, Mark Wilkinson wrote:

> Okay, I have thought about this now, and discussed it with Eddie and Ben
> for a while.  I can see the way Martin is thinking, and it is certainly
> different from how I have always "seen" MOBY Objects.
> 
> Here's how the argument goes:
> 
> Martin:  MOBY Objects are just syntax, so there is no such thing as a
> "core" data-type for MOBY objects.  They don't have an "essense"!
> Objects are just "stuff"
> 
> Mark:  Yes they do!  The ISA hierarchy implies that an object has an
> "essense".  There are two ways to construct an AnnotatedFASTA object:
> It can either inherit from FASTA, and add an Annotation, or it can
> inherit from Annotation and add a FASTA.  Under the new object creation
> system, the syntactic outcome of these paths is identical, but they are
> "qualitatively" different.... it is nonsensical to "FASTA an
> Annotation", but it is sensible to "Annotate a FASTA".  The "essense" of
> the object is the FASTA sequence it contains, based on its ISA
> hierarchy.  The problem we have now is that, without querying the
> ontology and traversing back to the most primitive object, I cannot know
> which contained object represents the "essense" of my in-hand object
> (i.e. the thing that I would MOST want to represent if I had to
> represent that object in a hurry).
> 
> It also causes a problem with rendering... a FASTA sequence and a
> Genbank record are formatted strings.  What happens now that we have
> contained primitives is that you have to ask the parent tag about its
> type before you know how to render the child tag. I can't render a
> string containing a genbank record without paying attention to its
> formatting, but I don't know that a String is a formatted string until I
> ask the parent GenBankFlatfile object what it is...  :-/
> 
>   >>sigh<<
> 
> I dunno... the world just got much harder.
> 
> M
> 
> 
> 
> 
> On Wed, 2005-11-02 at 17:51 -0800, Mark Wilkinson wrote:
> > Hi Martin,
> > 
> > I love using you as the bellweather for my own thoughts :-)  You are 
> > seldom wrong, and delight in telling me when I am! ;-)
> > 
> > I think I am starting to see your point - I had not been thinking of the 
> > object ontology in quite that way, but you've got me reconsidering now.  
> > I'm going to bring it up for discussion at my lab meeting tomorrow and 
> > try to hash-out with the local guys exactly why I feel the way I do, and 
> > whether or not there are any serious adverse effects to thinking about 
> > it your way.  It certainly makes things easier to do it your way, though 
> > I wish now that we had had this discussion while Eddie was writing the 
> > database update script because I would have changed the way we 
> > approached it...
> > 
> > Oh well...  I guess that's why MOBY is still considered a research 
> > project ;-)
> > 
> > Thanks for being so forthright in your opinons - you are an extremely 
> > valuable asset to the project!
> > 
> > M
> > 
> > 
> > 
> > _______________________________________________
> > MOBY-dev mailing list
> > MOBY-dev at biomoby.org
> > http://www.biomoby.org/mailman/listinfo/moby-dev
> 

-- 

Andrew Farmer
adf at ncgr.org
(505) 995-4464
Database Administrator/Software Developer
National Center for Genome Resources

---
"To live in the presence of great truths and eternal laws,
to be led by permanent ideals-
that is what keeps a man patient when the world ignores him,
and calm and unspoiled when the world praises him."
-Balzac
---