Re: comments on http://www.w3.org/TR/2002/WD-webont-req-20020307

Marton,

Thanks for you comments, please see my responses below.

Marton Trencseni wrote:
> 
> 2.1 web portals
> "academic papers are written by one or more authors, which are people;
> people have surnames and given names and affiliations, which are
> organizations"
> affiliations are not organizations, they are relations between people
> ang organization. if you would be creating an ontology for these terms,
> you would define people, organizations, and a relationship called
> affiliated, whose domain is people, range is organization.

Yes, the sentence could be a little more clear.

> 2.2 multimedia collections
> "Ontologies can be used to provide semantic annotations for collections
> of images, audio, or other non-textual objects. These annotations can
> support both indexing and search."
> what exactly do you mean by indexing vs. searching? indexing as in
> determining what directory to list a certain site (piece of multimedia)
> under? it seems to me, that one of the benefits of the "web of
> ontologies" ("ontology of the web"?) would be, that such people
> searching wouldn't have to rely on such directories.

Indexing could be thought of as classifying resources using terms from
the ontology and adding meta-data about them using properties from the
ontology. Searching is the process of matching a query with these
annotations in order to find relevant information.

> 2.3 corporate web site management
> "Furthermore, a parametric search is often more useful than a keyword
> search with taxonomies."
> what exactly do you mean by parametric search?

A parametric search is one that has one or more parameters. For example,
one might execute a search for a movie where "director=Lucas and
releaseDate < 1980." We'll have to add an explanation to the document.

> the use cases, in my opinion, give "localized" examples. by localized, i
> mean that the ontologies only need to describe a limited domain. in
> section 2.1, the first sentence states "Web portals are web sites that
> collect information on a common topic." in section 2.2, where the
> example give is multimedia collections, the domain is again quite
> limited. i don't see how this document covers the -problems- involved
> with defining ontologies, or relationsships between ontologies that are
> "powerful" enough to describe any website available on the Internet,
> although it is stated as a design goal: 3.3 ontology interoperability.

These use cases don't necessarily need to be localized. A portal could
serve multiple communities with some common interests, and might require
different ontologies to serve the differences between these communities.
A multimedia collection may have a common ontology for describing the
media, but may use any number of ontologies to describe the content of
the media. A multimedia collection fo scenes from Western movies should
probably also use an ontology that describes cowboys, horses, saloons,
gunfights etc. A collection of clips from country-western songs might
also use the same Old West ontology, but might also have other
ontologies that deal with pickup trucks, hunting dogs, or divorces.

For how the language is intended to support ontology interoperability,
look at the requirements and objectives that state that as a motivation.
Keep in mind, that we don't propose to completely solve this problem
with this version of the language. We intend to provide some features
that allow partial interoperability such as establishing equivalence
between terms in different ontologies and using terms from other
ontologies to form complex descriptions of terms in new ontologies.

> marton trencseni

Jeff Heflin

Received on Tuesday, 2 April 2002 11:34:48 UTC