Re: Proposal for use of labels in Manchester Syntax ISSUE-146, ACTION-247

From: Alan Ruttenberg <alanruttenberg@gmail.com>
Subject: Re: Proposal for use of labels in Manchester Syntax ISSUE-146, ACTION-247
Date: Wed, 4 Feb 2009 04:13:23 -0800

[I moved the following announcement to a more prominent position in the
 document.] 
> (In case there is any doubt, speaking as WG member, not chair)

>>> I honestly don't see the point of the format without this.
>>
>> Hmm.  Given that the format is already in use, I don't see how this
>> point of view can be sustained.
> 
> Here is how I see the situation. The predominant use of the Manchester
> syntax currently is editing within protege. 

I expect that this is the case.

> Within Protege what I call
> quotedLabels are consistently available for input and display. 

Good for Protege.

> One of
> the largest communities that currently uses OWL is the biomedical
> ontology community. Within that community it is considered best
> practice to use opaque URIs as identifiers (e.g. OBO, MeSH, UMLS).

This may or may not be a good idea.

> People used to working with those ontologies in Protege uniformly work
> in views that use labels. 

Fine.

> The Manchester Syntax, as solely an exchange
> syntax, offers no additional benefit because there are other viable
> exchange syntaxes. 

Agreed.  But who is selling the Manchester Syntax solely as an exchange
syntax?  I certainly am not.

> In order that it be worth having another syntax, it
> must bring some additional value. 

Sure.  

The Manchester Syntax provides a compact readable syntax for use in the
Primer.  Is this not additional value?

> Historically that value has been
> that it is human readable and editable. 

I do not believe that the major historical value for the Manchester
Syntax is that text whole-ontology documents written in the Manchester
Syntax are readable and editable.  Instead, I believe that the major
historical value for the Manchester Syntax is that complex descriptions
can be rendered in a readable and editable manner in it.

> It isn't if it can't be edited
> using human readable labels for entities.

I'm not arguing that having readable identifiers is bad.

However, I don't see that the additional costs are justified to add a
new kind of construct to the Manchester Syntax to allow ontologies 
exchanged in the Manchester Syntax to have a different kind of
identifier.   Among other things, I worry about ambiguity and unintended
effects of textual changes to an ontology document.

After all, if communities want to 
1/ use URIs that cannot be nicely abbreviated
and
2/ exchange ontologies in the Manchester Syntax in a form where the use
   of an object has a more intuitive label
they can always put these labels in comments.

Any other use appears to be better handled as intra-tool support.

> -Alan

Peter F. Patel-Schneider
Bell Labs Research

Received on Wednesday, 4 February 2009 12:58:29 UTC