Re: predicates as verbs

Paola/Henry,

I think that you are speaking about slightly different things. On the one
hand, Paola proposes a good practice to define the metamodel as triples. By
metamodel, I mean the relation between different objects.  For instance,
<Person, writes, email>. In OWL-DL, this would be called the TBox. On the
other hand, Henry states that this proposal is not always applicable. There
are in my views two exceptions to Paola's rule:

   1. to define attributes of a concept (datatype properties in OWL).
   2. to define data as triples (as Henry pointed out)

The latter is becomes even more prominent since the start of the LOD work,
where data is represented as RDF triples. Furthermore, the LOD work doesn't
require an ontology to be defined for the data to be made available.

I hope this helps.

Quentin

On 1 July 2010 17:31, Paola Di Maio <paola.dimaio@gmail.com> wrote:

> Henry
>
>
>  I am not trying to convince you or anyone of anything, I am reminding you
> the way it is
>
> the practice of data modelling is established, and there are a few, key
> fundamental rules
> that make things work, afaik the predicate as relation is one of them
> (unless you can refer me to different practices)
>
> relations need to be verbs if you want the data model to be sound in an
> economical way
> and possibly within my lifetime
>
>
> of course, everything else is possible too, you ll just have to find the
> time to make it work
>
> let me know when you do?
>
>
> P
>
>
>
>
>
>
> On Thu, Jul 1, 2010 at 4:21 PM, Henry Story <henry.story@gmail.com> wrote:
>
>> relations need not be verbs at all. Eg: +
>>
>> (2 2) plus 4.
>>
>> Rules such as the one you mention are just useful guides to help people
>> get going.
>> Verbs are actions and actions are events, and events can also be named
>>
>> 2010Olymbpics contains springing20234 . spinging20234 speed 24s .
>>
>> or something like that.
>>
>> Henry
>>
>> On 1 Jul 2010, at 17:13, Paola Di Maio wrote:
>>
>> > I am not following close enough the discussion on subjects as literals ,
>> and
>> > whtere they contribute
>> > to the awkwardness of RDF
>> >
>> > One thing I remember finding it disorienting though, is that there is no
>> > rule that a predicate must be a verb
>> > (when I was looking at triple as if it were a subject- predicate- object
>> > model )
>> >
>> > In standard modelling practices  (E/R modeling) the relations tend to be
>> > verbs
>> >
>> >
>> > CF PAGE 196
>> >
>> http://books.google.co.uk/books?id=1ftCwG4K-HEC&lpg=PA196&ots=FBX42Ms9Ys&dq=ENTITY%20RELATIONSHIP%20MODEL%20SENTENCE%20VERB%20RULE&pg=PA196#v=onepage&q&f=false
>> >
>> > I did bring this up on another list, and the engineers thought it would
>> be
>> > good practice to
>> > restrict predicates to verbs for obvious reasons (obvious to anyone who
>> does
>> > data models)
>> >
>> > Not sure how that would play if RDF is shown as  EAV (entity attribute
>> > value) rather thant SPO (subject predicate object)
>> >
>> > Just thought I d mention this, in case someone wants to fix RDF thats
>> the
>> > first crack I spottend a while back
>> >
>> > cheers
>> > PDM
>>
>>
>