REQDOC: Lynn's secret comments (LONG)

Gee, I really hadn't meant to be keeping any secrets!  I just thought that
my comments were long, wordy, unedited (so as to be both occasionally
impolite and not very well organized or user-friendly), and not entirely
relevant to the group as a whole.  But, for the die hards, they're included
here.  (I still have not taken the time to edit them for public
consumption/readability, so take them with a grain of salt or three.)




Lynn Andrea Stein wrote:

> Jeff, I'm sorry these are so late.  I've tried to keep them very
> specific and constructive.  Also, this isn't the whole document but is
> likely to be all I can get you in time.
>
> Lynn
>
> These comments were made on the Feb 7 2002 draft.  I have not had time
> to cross-check against the draft of 20 Feb.
>
> Section 1
>
> Replace "This document collects..." (para. 2)  and para. 3 with the
> following two para.s
>
> This document is intended to motivate the need for a Web Ontology
> language as well as the goals and subsequent undertakings of this
> working group.  It explains *what* a Web Ontology language should
> accomplish and *why*.  In later documents, we intend to describe a
> specific Web Ontology language (which we will call Owl) and to explain
> how Owl accomplishes the requirements and goals set forth here.
>
> This document specifies goals, requirements, and desired scenarios on
> which the Web Ontology Working Group has reached agreement.  It also
> descxribes other features that may or may not be included in our final
> product, but do not represent the current consensus of the working
> group.  Features in this latter category are noted as open issues.
>
> In para. 5, replace "a couple of months after" with "within a few months
> of"; also add a bullet:
>   o currently extant proposals for such a Web Ontology language as well
> as other proposals that mayb e made or refined within this interval.
>
> Section 1.1
>
> To replace part of first paragraph:
>
> There is a substantial history of this kind of work within the
> artificial intelligence and knowledge representation communities (and,
> indeed, the philosophy literature dating back millenia).  The WebOnt
> Working Group has undertakes its task with clear sight of this
> background literature but also with a firm understanding of the
> fundamental nature of the World Wide Web -- distributed, decentralized,
> and huge -- and of the ways in which this historical research will need
> to be adapted to make it pragmatically useful within the web context.
>
> (I think it's important to put in a historical footnote here but also to
> point out that that work is not directly applicable without
> modification.)
>
> Section 2
>
> Expand intro paragraph to include:
>   understanding what is needed in each case, what we want to acheive,
> and how existing World Wide Web technologies fall short of these goals.
>
> Overall comment on use cases:  It would be good to end each use case
> with requirements OR (perhaps better) to shift all of the requirements
> to a later section.  The lack of parallelism is disturbing, though.
>
> Section 2.1
>
> Replace last sentence of first para with
>   Each web portal allows individuals who are interested in the topic to
> receive news, find and talk to one another, build a community, and find
> links to other web resources of common interest.   [delete the community
> of and change that to who and add build a community]
>
> Next para:  providers'  (add apostrophe after s in providers) n Same
> modification in next paragraph (providers' annotating)
>
> Next para:  may not provide the community....the content that ITS
> MEMBERS require...  Also add an example as this does not seem
> sufficiently evocative.
>
> Section 2.2
>
> I generally found this example less general and straightforward than
> many of the others.  Perhaps it could be generalized slightly to include
> audio and other non-text web objects (though the examples could remain
> specific).  I think that the issues are largely similar.
>
> Ontologies can be used to provide semantic annotation ofr collections of
> images, audio, or other non-textual web objects.  (Rest of para. same,
> except "...can describe these nontextual objects in different ways..."
> and "...retrieval of nontextual objects without requiring
> domain-specific search tools".)
>
> Second para then starts
>
> An ontology for nontextual objects should have the following features.
> ...utilize the part structure of the nontextual objects depicted...
> ("utilize part structure" doesn't work as a phrase for me.  Add THE?)
>
> The example is *extremely* specific and perhaps as a consequence less
> evocative.  It also speaks more to the domain of antique furniture than
> to the domain of image (or nontextual object) collections.  An example
> such as "Arctic Fjord" or "Picasso's Blue Period" might be more
> generally accessible.  But if you are going to stick with these
> examples, it would be a good idea to provide an example to support the
> taxonomy claim and the part structure claim.
>
> Also typo:  the a "chest of drawers"
>
> And finally, a chest of drawers is simply NOT typically made of
> mahogany, though perhaps a Late Georgian chest would be.
>
> Section 2.3
>
> Why/in what sense is this "corporate" web site management?  Any large
> organization, no?  I guess this one doesn't feel like it hangs together
> so well either.  It certainly begins in a textually awkward way.  (What
> is the point?  I suppose I'm still not sure even after having read it.)
> Again, I think that some of the issues might be the overly specific
> language.  (Also specific issue:  large corporation's  -- add
> apostrophe)
>
> E.g., the salesperson example strikes me as:
>
> Different members of the organization use different vocabularies.  For
> example, ...  (And the phrase "So keyword sarch will often be
> inadequate." is not a sentence, it's a fragment, not to mention an
> awkward one.)
>
> and the technical person e.g. looks like
>
> identifying experts within the organization
>
> or
>
> Q+A/FAQ or a knowledge repository
>
> And the final example seems like the need to
>
> assemble case studies of past experience within the organization
>
> I find the whole case study eg, from "Consider a past project..."
> through "..Furthermore, a parametric search" to be reasonably
> incomprehensible.
>
> Also, the para. that starts It is important to note   is not really a
> use case item; it doesn't belong in this section at all, though it may
> belong elsewhere in this document.
>
> I'd probably pull the requirements from the specific use case into a
> later section of the document.
>
> Section 2.4
>
> Again, make general.  There's no way that this is about aerospace
> engineering documentation.
>
> The task involved in this use case is to classify and crosslink a large
> body of engineering documentation.  This documentation falls into
> several different types:  ...
>
> Last sentence of secont paragraph is awkward.  Physical objects and
> their ontologies are developed in tandem?
>
> "effectivity" is not a word.  Effectiveness?   Also, "we may need to
> know, e.g., for a given aircraft...."   This whole paragraph is about
> retrospective analysis, e.g. in the case of an airplane accident.  (It's
> more general than aerospace...)
>
> Same issue with respect to the requirements in this section; I'd move
> them all later.
>
> Section 2.5
>
> An *evening* planner?  Maybe an itinterary planner or an activities or
> social planner.  (Similar change of the word "evening" throughout)  This
> example seems to mix specifics in a funny way (e.g., sf.yahoo.com?!?)  I
> think it is really about planning an itinerary.
>
> Section 2.7
>
> Add a summary of requirements.  This would be a reasonable place to
> collect the requirements from the individual tasks above.
>
> Section 3
>
> 3.1 Shared ontologies
> Justification
>
> OK, but the first paragraph is an explanation, not a
> justification....this isn't WHY we need shared ontologies.  Also, how do
> we enforce the alleged commitments you're describing here.
>
> A better justification:
>
> (1) need the benefits of communicating in terms of shared definitions
> (2) need to minimize the work for "near miss" ontologies, i.e., when
> there's an almost-appropriate ontology
>
> (In your text:  90% of what IT needS, not THEY NEED)
>
> RDF support:  Combining the schema's  (insert ')
> Also RDF isn't unclear on anything; The RDF spec might be.
>
> 3.2 Ontology evolution
> Justification:
>
> errors in prior versions, BECAUSE a new way..., or BECAUSE
> reality...(e.g., BY the additon...)
>
> ...without changing any axioms:  How would I know?  How would this be
> enforced?  What does it mean to change an intended meaning, anyway.
> (What effect does the intention of the author have?!?)
>
> If you keep the text, break the sentence after axioms (PERIOD)  (new
> sentence) Thus, determining...
>
> 3.3 Inconsistency detection
>
> ...it should be possible to detect....
>
> computationally??!??  But isn't this the halting problem?  Heck,
> propositionally it's SAT, which is pretty nasty.  This is decidability!!
>
> 3.4 Ontology interoperability
>
> no comma after different ontologies
>
> Justification:
>
> Realistically, this will happen on the web.  OWL ought to support it.
>
> 4 Requirements
>
> Did you collect all of the requirements listed in the use cases?
>
> Say explicitly at the end of the introductory paragraph:
>
> The motivations listed here are references back to the use cases of
> section 2 where at least one motivating example for each of these
> requirements is presented.
>
> More tomorrow if I can, but this should be enough for you :o)
>
> p.s. not sent to webont-wg cause I didn't edit this note carefully, but
> if you want to circulate it by all means do so.

Received on Friday, 22 February 2002 14:50:23 UTC