Re: Lists of Values - Proofread version - supercedes all previous

Hi all,

Here are my comments on the draft (based on v5).  They naturally 
reflect my particular prejudices.

1.  Comments

Firstly, I like this document.  It is clear and useful.  It definitely 
addresses a real need.  Kudos to all involved.

However, it seems to me that an important real-world feature of 
deployed systems will be the existence of more than one ontology (or 
system of ontologies).  This has been discussed previously, but I think 
it is more important now that we are seeing existing, deployed 
implementations.  They will require mappings between ontologies in 
order to ensure interoperability and the ability to combine them.

Do we need to address this issue in this document?  I suggest that we 
do.  The reason is that the two patterns suggested in the draft are not 
equal in terms of the mapping requirement.  Only Pattern 1 has the 
features of alternative partitionings and subpartitionings, which I 
think would be needed.

Pattern 2 seems to exist primarily to ease implementation, a goal I 
strongly support!  However, Pattern 2 only eases implementation if the 
underlying store has certain properties.  I don't think we can presume 
(or should even suggest) technologies to use when implementing 
recommendations.  However, we do want to ensure that recommendations 
may be readily implemented and Pattern 1 addresses this issue.

My apologies if I appear to be taking sides on an issue long resolved.  
I am new to the WG and reviewed the draft in relative isolation.  If 
this issue has been discussed and a consensus reached, please feel free 
to ignore the above.

The "General Issue" Section notes that data typing will not be 
considered in this paper and a supplement may be issued later.  I would 
like to strongly encourage that discussion and an issuing of a 
supplement.  Simon Raboczi, on the DAWG, has made some suggestions in 
relation to generalizing data types with resources.  His ideas could 
have substantial impact on type enumeration in OWL.


2.  Typos and Other Nit Picks

Section "General Issue":  othrwise -> otherwise

Section "Use Case examples":
   - 'qualities body type' -> 'qualities of body type'?

Section "Pattern 1: Values as subclasses partitioning a quality":
   - 'Johns_health' appears in mixed font in the HTML version.
   - Figure 1 suggests that a Healthy_Person has_health_status 
Poor_health_value!  Surely that is supposed to be Good_health_value...

Section "Considerations using Pattern 1":
   - 'partitioning a continuous quality space' -> 'partitioning of a 
continuous quality space'

Section "Pattern 2: Values as individuals whose enumeration is 
equivalent to the quality":
   - The last sentence in the first paragraph ('If we did not 
include...') is in mixed font in the HTML version.

Section "Representation for Pattern 2":
   - ':good_healthH' -> ':good_health'

Section "Notes":
   - 'and and individual value' -> 'and an individual value'

Section "Acknowledgements":
   - 'Protege-OWl plugin' -> 'Protege OWL plugin'

Regards,
Dave
--
David Wood
CTO, Tucana Technologies, Inc.
http://www.tucanatech.com
--------------------> Information Fusion. Tucana. <--------------------

On Jun 25, 2004, at 04:29, Alan Rector wrote:

> Tom - and All
>
> The proofreaders got back to me after the Telcon.  Here is a corrected 
> version which should be the one to go out for comment.  Apologies for 
> my inability to do this without help and any inconvenience and the 
> various circumstances that meant it couldn't be done sooner.
>
> Regards
>
> Alan
>
>
> --
> Alan L Rector
> Professor of Medical Informatics
> Department of Computer Science
> University of Manchester
> Manchester M13 9PL, UK
> TEL: +44-161-275-6188/6149/7183
> FAX: +44-161-275-6236/6204
> Room: 2.88a, Kilburn Building
> email: rector@cs.man.ac.uk
> web: www.cs.man.ac.uk/mig
>         www.opengalen.org
>         www.clinical-escience.org
>
> <Lists-of-values-5.zip>

Received on Wednesday, 7 July 2004 14:50:16 UTC