Talk:Primer

From OWL
Revision as of 22:43, 13 April 2008 by BijanParsia (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

Review from Jie Bao

Overall impression

The document has done a great job in providing a consistent running example of family relations, which, if compared with the wine and food example in the old OWL Guide, is more easily comprehensible to the general audience. However, as the majority of the general audience will be web end users (as the "W" in OWL stands for Web),

Review comment from BijanParsia 16:02, 13 April 2008 (EDT)
There is no evidence for that, nor are "web end users" by any means a monolithic bunch.

not only tool builders,

Review comment from BijanParsia 16:02, 13 April 2008 (EDT)
The primer is not at all targeted to tool builders. Quite the contrary. Evidence otherwise so we can correct that is welcome.

some reorganization and rewording may be needed to make the document even more comprehensible to readers.

Some parts of the Primer have a strong favor as a tutorial of a knowledge representation language. Some parts of the discussion, e.g., on open world semantics, might be too complicated and too detailed for usual users.

Review comment from BijanParsia 16:02, 13 April 2008 (EDT)
I don't know who you think the usual users are. In my experience, not explaining these things causes more problems than not.

It seems the current draft assumes the familiarity of basic notions like classes, properties and individuals of readers, which is not always true for end users. More explanations and examples will be more valuable, especially for the "Basic Notions" section. Previously, in OWL Guide, there was an incremental introduction on basic elements of an ontology, which is much easier to understand by non-expert (I believe). So it is nice to keep that pattern in the Primer.

Review comment from BijanParsia 16:02, 13 April 2008 (EDT)
I personally don't see that pattern in the guide. It starts with a heavy discussion of namespaces, after all. However, we do intend to be as clear and bootstrapping as possible. The question isn't the goal, it's how do we best achieve it.

Several important constructs are missing in the Primer, e.g. objectAllValuesFrom, objectSomeValuesFrom (as Uli pointed out) and negativeObjectPropertyAssertion.

Review comment from BijanParsia 16:02, 13 April 2008 (EDT)
It's not clear that discussing every construct is the best tactic. My hope is that we can give insight that will make using the reference easier. An exhaustive discussion of all features will be exhausting.

Detailed Review

Section 1 Introduction

Paragraph 1

  • "OWL is a logical language" - is it true for all species including OWL FULL?
Review comment from BijanParsia 07:08, 8 April 2008 (EDT)
Yes.
  • TYPO: OWL groups information into ontologies in the form of documents that can be stored and transmitted (in) on the World Wide Web in the same way (that) as data and other kinds of information are. and that Ontologies in OWL can be completely (what this mean? a weaker word is preferred - Jie) and effectively processed by tools that extract the information implicit in an ontology.
Review comment from BijanParsia 07:08, 8 April 2008 (EDT)
Feel free to correct typos as you find them directly. This is a wiki after all.
  • "Ontology" should be explained. Most of web users who need the Primer will not know what is an ontology.
Review comment from BijanParsia 07:08, 8 April 2008 (EDT)
I'm not sure what to say. Explicit definitions have value, but so do implicit ones. I'll ponder more.

Section 1 overall

  • AT END OF THE SECTION: should mention briefly the history of OWL and the motivation why OWL 1.1 is made, and say that by default "OWL" in the Primer refers to "OWL 1.1".
Review comment from BijanParsia 07:08, 8 April 2008 (EDT)
See the email exchange between me and Deb. I'm not sure *at all* that a history is a good idea at this point (perhaps at all). Saying that OWL means OWL 2 is a good idea and I'll add it.

Section 2 Orientation

Section 2 overall

This section is very technical. However, it may scare off many users who have no KR background (which I believe will be the majority of users). I strongly prefer to hide the discussion on open world/close world discussion as much as possible from Orientation, as it will be further discussed in Advanced Notions.

Review comment from BijanParsia 16:14, 13 April 2008 (EDT)
This has been discussed extensively. It was never the intention that any novice slog through all this if it wouldn't help them. After all, they are explicitly mention as for people with such and such background. However, it seems we'll be moving these anyway.

Two major ways of using OWL, Terminology development and management and Conceptual Modeling, are mentioned. As a Web ontology language, OWL's primary goal is surely more than them,

Review comment from BijanParsia 16:14, 13 April 2008 (EDT)
These are *two* major ways, not all major ways.

and probably, the two points mentioned here are not every day practice of end users.

Review comment from BijanParsia 16:14, 13 April 2008 (EDT)
I don't know what end users you work with, but these are extremely common. I would say, in fact, that they are dominant. Moreover the point isn't to document every way but to provide some characteristic uses that show interested advantages of OWL

Also, they are ways of how OWL can be used, not applications that OWL can support. For example, in the OWL Guide, a Wine Portal example is shown to demonstrate what OWL can serve web information providers and consumers. From the point of view of an end user, it will be more interesting to show how OWL can help, e.g., in information integration (e.g. semantic portal), semantic search, etc.

Review comment from BijanParsia 16:14, 13 April 2008 (EDT)
I don't find the wine portal remotely compelling. It doesn't illustrate any particular strengths of OWL and it's all to easy to see how one would hard code it. As for the other examples, what do you think terminologies and conceptual models *do*? They help with info integration, search, etc.

Organization

  • Difference to RDF,... OO should be 2.1; 2.1 and 2.2 should be 2.2.1 and 2.2.2

Difference from RDF - OOP

  • XML: Why not reuse OWL Guide's wording as pointing out XML is a message format language and OWL as a KR langugage?
Review comment from BijanParsia 16:14, 13 April 2008 (EDT)
Because XML isn't a message format language, for one? As a member of the XML community, I always found that line ridiculous. And how will saying that OWL is a KR language help all those WEB end users you claim will be turned off by a KR orientation?
  • Database: Completeness/incompleteness should not be used as the main SELLING point for the difference between DB and OWL. I believe stronger modeling ability of OWL is a better selling point to end user. Incompleteness is rather a technical detail. I suggest to merge and shorten the first 2 paragraphs and put it as the last paragraph of the subsection.
    • It is not clear why no support for integrity constraint in OWL is an advantage? Actually, many user may expect this feature.
  • OOP: "Object-oriented programming (OOP) also has object-centered modeling characteristics" - why "also"? OWL is not OO in the traditional sense.
Review comment from BijanParsia 16:14, 13 April 2008 (EDT)
It depends on what you mean by traditional sense. In any case, it's perfectly true that OWL and OOP languages are object centered.
    • "Similarly OOP classes are much less expressive than OWL classes." - reference?

2.1 (Teminology)Terminology development and management

  • Typo (Para 1): Predefined terminologies may also be used at data entry either to (categlog) catalog the entry...
Review comment from BijanParsia 16:14, 13 April 2008 (EDT)
Please just correct typos directly

2.2 Conceptual Modeling

This subsection is rather academic. I'm afraid few end user (if OWL is intended to be a WEB ontology language) really need to do the job of Conceptual Modeling.

Review comment from BijanParsia 16:14, 13 April 2008 (EDT)
You keep saying what you think they won't do, but you don't say what you think they WILL do. CM is almost certainly what most people, esp. dabblers, do with OWL. Think UML.

Section 3 Basic Notions

Organization

  • Should be further divided into subsection and subsubsections.
  • Basic constructs like "and", "or", "not" should be introduced.


Detailed

  • (para 2) "The RDF/XML syntax for OWL is just RDF/XML,..." - what that means?
  • (para 2) [typo] "can be viewed (an)in any of the four different syntaxes"
  • f prefix is suddenly used without introduction (only later in Section 5) - better to explain it here.
  • The basic notions of individual, class and property should be better introduced in an incremental manner.
  • Naming conventions should be said for individuals, Class (Capital initial of each word), Property (lowercase first + capital initial from the 2nd word ). There is a global abuse of property names throughout, e.g. use "son" as the short of "hasSon". It is misleading somehow, as many user will be confused whether "son" is used as a class (which follows from intuition) or a property (which is what is in the document). Better to the consistent in naming.
  • TYPO: "OWL (has) is a rich language for defining classes in terms of other classes, but also in terms of the relations its (of which? -Jie ) instances may or must have to other individuals."
  • Wording: "this would provide exactly the same information to OWL, and OWL reasoners can determine this."
  • Wording: "It may seem that there is a circularity in defining Parent as people with at least one child and also making it be the domain of child. In OWL, however, there is no problem. The two bits of information are simply different ways of saying the same thing." - this point is not apparent to common users. To understand this, user are required to have a quite bit of knowledge of DL. Also, it may not be necessary to mention it in the Primer.

Section 4 Advanced Notions

Overall

This section is heavily focused on the open world semantics and the ability to model incomplete knowledge. It is indeed an important charactertistics of OWL. However, it might not be what an end user expects in such a section. I would like to see this section as a replacement of the section 3.3 (Property Characteristrics) and 3.4 (Property Restrictions) of the Guide. That is, it will give a guide of user on how to use some constructs.

If open world semantics will be a serious concern, I would like to suggest to have a devoted section that is separated from basic notions and advanced notions on it. Section 3 and 4 should focus on how to use constructs, from the easiest case to somehow more complicated usage, for users who even have no any idea of what is "incomplete modeling" to use some power of OWL.

This section also needs to be broke into more subsections.

Details

  • "In OWL we can have transitive properties, i.e., properties like hasAncestor, which also is a generalization of the inverse of the hasChild property, and is also irreflexive." - it may lead to confusion as I have had. How about to first define hasDescendent as the inverse of hasAncestor, and then define hasChild as a subproperty of hasDescendent?

Section 5 Ontology Management

Organization

This section addresses mainly three sets of constructs of OWL

  • Housekeeping constructs like Ontology url, namespace and importing
  • Ontology mapping constructs between individuals, classes and properties
  • Annotation and declaration

I see the mapping constructs are more close to a feature of modeling, instead of part of ontology management. In the old Guide, there is a devoted section on Ontology Mapping (section 4) with extensive examples. We may break the current section 5 into three subsections "ontology management", "ontology mapping" and "annotation" with a new section name "Manage and Connect Ontologies".


details

  • "An OWL annotation simply associates property-value pairs with parts of an ontology, or the entire ontology itself." - may need rewording. More words are needed to explain what are "property-value pairs" (may after the example) and what are "parts of an ontology" (e.g., a class or a property-value pair)

Section 9 OWL Species

  • Para 2 talks about decidability - this needs some explaination e.g. all entailments are guranteed to be computed in finite time.
  • If the WG decides to keep OWL-Lite as it was, a new paragraph shall be added about OWL-Lite.
  • I agree that we should mention its relationship to the others. I think we currently can do that rather simply just to say it is a subset of OWL DL. Deborah McGuinness

Section 11 OWL features

Shall we mention the difference between OWL 1.0 and OWL 1.1? Some users may still wish to keep inside OWL 1.0 and we should allow them to do so.

Section 13 References

Don't you think that would be rather confusing? RinkeHoekstra 13:57, 19 March 2008 (EDT)

I think it is important to reference the Guide since this is the document that this document is meant to replace. I think it would be confusing not to reference it. Deborah McGuinness

Reference