Review of PROV-O - ISSUE-336

Hi All,

I reviewed PROV-O at
http://dvcs.w3.org/hg/prov/raw-file/tip/ontology/spwd/for-internal-wg-review/prov-o.html

First, great job. This is a nice product to work with. Especially, the
connection between the owl and the document.  Good job Tim and
everyone.

I have formatted my comments below in markdown so it's best to read in
a markdown viewer.

Thanks
Paul

#Responses to Editor's Questions#

* Does the HTML file provide an adequate overview of the OWL design elements?
    * Yes.
* Do the different organizations of PROV-O HTML and DM complement each
other, or is it distracting?
    * I think it's fine as it is. It seems that it takes a different
approach which is fine. I think one should avoid introducing the dm
organization into the prov-o.
* Would any additional comments (or attributes) help you read the
cross reference list in PROV-O HTML?
    * I wonder if one example for each of the constructs should appear
in the cross reference list.
    * A link to the prov-dm document would be good.
* Are the comments within the OWL file adequate to familiarize with
the structure? If not, what kinds of comments would help?
    * Their great. I've asked for some clarification and additions below.
* Should the OWL file contain any links to documentation (e.g., to the
DM, to examples, etc.)?
    * Yes.. see my comments on the annotations of the ontology below.
* Can the document be released as a next public working draft? If no,
what are the blocking issues?
   * Yes. Great work. It's time for the community to see what they think.

# Comments on Document #

## Status of the Document ##

* it would be good to see links to the provenance ontology
* There should be a brief description of the changes since the last
working draft.

## Introduction ##

* I think a link to the ontology itself should be very visible in the
introduction. e.g. "The ontology can be downloaded from here"
* The namespace should be more visible in the introduction.
* The following sentence is repetitive - "The PROV Ontology can be
used directly in a domain application, though many domain applications
may require specialization of PROV-O Classes and Properties for
representing domain-specific provenance information."
* OWL-RL is just mentioned. Does this need to be motivated?

## PROV-O at glance ##
* The classes and properties should be labelled as such in the boxes.
* It says "additional terms that can be used to relate classes in the
Starting Point category." - is it just used to relate starting point
classes? Maybe a revised sentence could be "additional terms that can
be used to augment provenance information expressed using constructs
from the Starting Points category"
* I wonder if collection classes and properties should become a
subsection or linked from the expanded classes and properties section.
In general, to me they are on par with these other notions and it
seems that we both over and understate their importance with a
separate section.

##The PROV-O Ontology Description##
* I really like the diagram!
* It would be good to put links to the definition of the properties as
you introduce them. Maybe actually drawing out these definitions as
bullet points ws would be good.
* In the example, can we change the organization from "civil action
group" to "external organization" or a "newspaper"
* The explanation of the example is hard to read because of the
examples in orange. A picture of the example would be good here.

##Expanded Terms##
* Not all the terms are introduced. The section kind of stops with the
example. I would suggest saying that either that this is just an
example or introduce more of the terms
* You talk about shortcuts but what this means is not clear. This
should either be dropped or expanded.

##Qualified Term##
* Great section. Really shows how to use qualification.
* The Qualified Pattern is introduced but I would expect the pattern
itself in its generic form to be introduced and then an example of its
application given.

##Collections##
* A diagram would be especially helpful here showing how each
collection is built up.
* I would emphasize that this is not to represent collections
generically but to represent how collections are constructed or
created.

##Cross reference##
* A very nice reference over all.
* I would remove "in PROV component", it's confusing.
* The notation should be introduced (e.g op and dp)
* Some of the classes and properties are missing definitions. I assume
this can be fixed by importing from prov-dm
* What is a property and what is a class should be denoted more
clearly. e.g. Class - prov:Activity
* In some places there are additional notes for example in prov:Derivation:
    > An instance of prov:Derivation provides additional descriptions
    > about the binary prov:wasDerivedFrom relation from some prov:Entity
    > to another prov:Entity. For example, :chewed_bubble_gum
    > prov:wasDerivedFrom :unwrapped_bubble_gum; prov:qualified
    > [ a prov:Derivation; prov:entity :unwrapped_bubble_gum; :foo :bar ].

    * These should be set-off some how. I assume these are "usage
notes" or something like that.
* There should be a link to the prov-dm definition.

#OWL Document#

##Annotations##
* It would be good to define your annotations in the comments to the
owl file. For example, I didn't first grasp the difference between
category and component in the annotations. This could be also given in
the comments of the annotation definitions.
* The class hierarchy is nicely balanced. However, there are some
things that stick out. For example, KeyValuePair, Membership and
Location all seem to be dangling. I wonder if we can collect these
somewhere.
    * Is Membership an Involvement?
* In the comments of Involvement should we note that this is used for
organization purposes?
* properties do not seem to have rdfs:label
* refs:label need to be defined with their language tag
* should we define the category and component with urls. Currently, we
are using keywords to define our vocabulary here. It seems that being
able to dereference these definitions would be good.
* The links to the dm are nice as annotations.
* __How do we integrate "strings" in the ontology?__ This is important
because we need good ways to integrate application data into the
ontology. I think there should be some patterns to adding these
strings and these should be clear. For example, we may mint new urls
for each entity, but for example, with a quote we may want to easily
add some text.
* It would be good to have back pointers to the ontology document
itself as annotation properties.
* Hmm… I wonder how in the ontology we can point people to the
involved set of properties, as these are actually the core but it's
not what one sees at the top-level





-- 
--
Dr. Paul Groth (p.t.groth@vu.nl)
http://www.few.vu.nl/~pgroth/
Assistant Professor
Knowledge Representation & Reasoning Group
Artificial Intelligence Section
Department of Computer Science
VU University Amsterdam

Received on Sunday, 8 April 2012 15:43:33 UTC