W3C home > Mailing lists > Public > www-rdf-interest@w3.org > January 2001

XRDF - an eXtensible Resource Description Framework

From: Reinhold Klapsing <Reinhold.Klapsing@uni-essen.de>
Date: Mon, 22 Jan 2001 19:11:40 +0100
Message-ID: <3A6C77DC.B708A699@uni-essen.de>
To: www-rdf-interest@w3.org
Dear RDF-IG members,

we invite you to have a glance at the discussion paper:
XRDF - an eXtensible Resource Description Framework

Accessible at:
- http://nestroy.wi-inf.uni-essen.de/rdf/xrdf/xrdf_v10.ps
- http://nestroy.wi-inf.uni-essen.de/rdf/xrdf/xrdf_v10.pdf
- http://nestroy.wi-inf.uni-essen.de/rdf/xrdf (XRDF-Home-Page)


This paper presents a nested triple model for
expressing relations found in the Web. The model
allows grouping of atoms and statements on
subject and object position. It preserves the
structural context in which resource are used.
Additionally, we propose a (pure) XML
serialization syntax and a graphical
representation which equivalently express the
formal concepts. On top of the basic structural
layer, semantic definitions and interpretations
can be layered. One such layer is presented.
Finally, the relation of this approach to RDF is
discussed and it is argued, that most of the
perceived deficiencies of RDF are non-issues in
the context of XRDF.
Below, you find a brief review of the key features and some general
remarks on our intentions.

Some key features are:
* A simple structural "model":
  In the following definition, the (infinite)
  alphabet A* will be used. A* will denote all
  possible instances of PCDATA in well-formed XML.

  We define the structure R recursively as an expression over A* as

  R ::= r  |  R,R  |  [[R],r,[R]]

  The terminals r denote elements of A*.  (you may call the possible
  structures "resources", but we tried to keep it free from
  interpretation even on this basic level)

* A straightforward XML sntax correspong to the structural model,
  <!ELEMENT statement (subject, predicate, object)>
  <!ELEMENT list      (statement | atom)+>
  <!ELEMENT atom      (#PCDATA)>
  <!ELEMENT subject   (atom|statement|list)>
  <!ELEMENT predicate (atom)>
  <!ELEMENT object    (atom|statement|list)>

* A graphical model directly corresponding to the structural model,
  allowing for grouping/sequencing.

* Some basic transformations are given, allowing to (de)reference
  representations, to "dissolve" n:m,1:m,n:1 relations etc. These
  basic tranformation can be used to embed the XRDF structures into a
  host formalism (such as FOL) (or, in other words: to transform XRDF
  expressions into "menaingful" expressions in other formalisms.

Some general remarks:
We tried to keep the structural model, the
synatactical and graphical representation and the basic
transformations as free from interpretation as possible. Some will
miss terms as "assertions" or "meaning/interpretation". This paper is
a "part I" that tries to offer a simple (yet powerful) recursive "data
model" with "positions" (based on triples again ;), a straightforward
syntax that allows to "build" deeply nested expressions with complex
(syntactical) structure (neat for "context"), and some basic
(structural) transformations tied to "predicates" (that are the
"things" in the middle of a triple). There is no semantics yet -
instead we tried to provide the ingredients that allows to plug the
XRDF stuff into suitable formalisms (by offering the tools that are
needed to transform structured expression into a different "language",
which may/should then be used to give meaning/interpretation to the
XRDF constructs. We do not think that fixing interpretations on this
level of langugae design is necessary or suitable - we feel that
different interpretations in different formalisms should be possible
easily. We hope that the discussion will show that there is much more
to say.

Please, allow one more word: we did not intend to "replace" RDF -- we
have a simple RDF-to-XRDF converter online and a "XRDF flatener" is
available in alpha version that allows, with a suitable set of
additional semantic rules, to convert XRDF to RDF (using reification,
position information, and dereferencing) -- instead, we thought that
it might be fruitful to discuss a somewhat "clear-cut" approach to
show which problems of RDF need to be tackled and how possible
solutions may look like.

 Thank you in advance,
      Wolfram, Reinhold, Eckhart
PS1: Any comments/questions/remarks are welcome. If
you think that the question/comment is not (yet) of public interests,
you may want to send it to rdf@nestroy.wi-inf.uni-essen.de only,
otherwise, you may also want to CC it to the RDF-IG. We will collect
all discussion on a Web page that will be accesible via

PS2: We wrote the initial version Oktober 2000. We decided to submit
it to the WWW10 conference -- mainly because, due to some unfortunate
personal circumstances, there would have been no possibility to follow
a possible discussion on the RDF-IG -- however, the paper was (and
still is) initially intended to be a contribution to the RDF-IG. In
the meantime, the time constraints have relaxed (so we are ready to
start the discussion now! ;) and we have received the comments from
the 3 reviewers. Two have been positive (7 and 6) (including
encouraging comments) and one was negative (3) (without further
comments) -- which is not enough to allow travelling to Hongkong but
instead gives more time to work on discussing and improving the stuff.
Received on Monday, 22 January 2001 13:11:42 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 15:07:34 UTC