Re: Is it a semantic problem?

Danny Ayers wrote -

> 
> What Tom said and...
> given the three samples that have come up, turning them into statements:
...

Let me follow my own advice and try to put Danny's three fragments into words, just to illustrate what I meant by the advice -

1) 
> <rdf:Description rdf:about="http://url.com">
>    <news:title>Some title Here</news:title>
>    <news:description>This is fun</news:description>
> </rdf:Description>

Translation -

"There is something identified by 'http://url.com', that has a news:title of 'Some title here', and a news:description of 'This is fun'."

2) 
> <item rdf:about="http://url.com">
>    <title>Some title Here</title>
>    <description>This is fun</description>
> </item>

Translation -
"There is something identified by 'http://url.com'.  It is an instance of an "item", and it has a title of 'Some title here', and a description of 'This is fun'."

In this translation, I am obviously ignoring the default namespace.   Note that Danny called the thing a "class', but you don't have to.  How do we know if an "item" is an rdfs class, or an OWL class, or something else?  We don't have to know.  You could add that information as another restriction on the thing, but you don't have to.  If you don't, you are not entitled to make inferences about  what kind of thing it is, unless you know by some other means that an "item" is indeed a class of some specified variety.

All you can infer here is that "item" is the rdfs:type of the subject.  You can check this using the W3C rdf validator, which will create a statement that says so, for example, 

	http://url.com 	http://www.w3.org/1999/02/22-rdf-syntax-ns#type 	http://example.com/item

3) 
> <rdf:Description rdf:about="http://url.com">
>    <news:item rdf:parseType="Resource">
>        <news:title>Some title Here</news:title>
>        <news:description>This is fun</news:description>
>    </news:item>
> </rdf:Description>

Translation -
"There is something identified by 'http://url/com', which has a news:item property, which in turn is something that has a news:title having a value of 'Some title here', and a news:description having a value 'This is fun'."

If we wanted to be more informative, we could add a statement that the bnode representing the news item has a rdfs:type of, say, news:Item (or whatever it should be).  However, it might be possible for the processer to infer that, depending on whether the property news:item has been defined with a range in rdfs or OWL.

Now that we have spoken all three sentences in natural language, we are in a better position to know if any of them express what we intended to express.  For example, do the title and description apply to the news item or to the thing itself?

Notice that the this particular question is not about modeling, but about that actual facts that we want to communicate.  If we decide that 3) is what we have in mind, then the matter comes up of how to model the facts.  Only after that should we think about the details of rdf-xml, which can take many forms and still describe the very same triples.

Cheers,

Tom P

Received on Monday, 20 June 2005 17:03:51 UTC