[AMF-talk] RePEc AMF Data Set for BKN

[Thomas Krichel]

  Frederick Giasson writes

> Now, lets ask a couple of questions vis-a-vis AMF itself.  First,
> the conversion should be quite straightforward since you already
> included some dc & dcterms (and other ontologies) logic in this data
> model.

  It's not realy important. There are so many ontologies
  around... 

> However I noticed that not real constraints were put on the use of
> adjectives and verbs to describe nouns. If we take a look at this
> section (and their description in the XSD file): "3.1.2: p/o to p/o
> verbs", I think I have a problem with the fact that "ispartof" can
> be used between two "person".

  Sure. It would not be making much sense. If you find such 
  a relation, you can raise an exception.

  One issue that came up when this was done is the semantics
  of "part" vs "member". I felt that it could be beneficial
  to make a distinction here. If A is a part of B, then when
  A goes away, so does B. If A is a member of B, then when
  A goes away, B may stay. In the end, I chose not to have this
  because it would just create too much confusion when to use
  one vs the other. 

> First question: are you aware of any person <--> person description
> that are linked by such a verb?

  No.

> Or is it only used to link two organization?

  Yes.

> This is where the mapping between AMF and the BIBO-BKN ontology will
> differ. In the mapping, the ispartof won't be used to describe any
> "person" noun.

  Why not?

> And if the converter encounter such a property that describes an
> organization, the relationship will be specialized and we will talk
> about "Affiliations" (different kind of affiliation exists; which
> can be used to describe these different facts). 


  I am not sure I follow this without knowing more about the BKN
  format.

> Otherwise, for the "text" noun, the "type" adjective can be omitted;
> if it is not defined for a "text" noun, then this text will be
> described using the higher class in the "document" hierarchy which
> will be bibo:Document. Otherwise, if this adjective is defined for a
> given "text", then we will use a subclass to refer to it.

  Sounds eminently sensible.


  Cheers,

  Thomas Krichel                    http://openlib.org/home/krichel
                                RePEc:per:1965-06-05:thomas_krichel
                                               skype: thomaskrichel