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

announce: New Tool for Comparing RDF Graphs (blindfold v0.1.1)

From: Sandro Hawke <sandro@w3.org>
Date: Thu, 04 Oct 2001 11:46:40 -0400
Message-Id: <200110041546.f94Fkfw10547@wadimousa.hawke.org>
To: www-rdf-interest@w3.org, www-rdf-logic@w3.org
I've implemented some RDF Graph comparison operations in blindfold (my
nascent (possibly inchoate) semantic web toolkit).  Brief documentation
and download instructions are at

      http://www.w3.org/2001/06/blindfold/

The code is probably alpha quality.  It aint perfect, but it does seem
to do the right thing.  It's most likely to work best in environments
like mine: Linux (RH7.1) with XSB 2.4 [1], but it's supposed to work in
all posix environments.

The problem I'm addressing here is: Do two RDF/XML files (or other RDF
graph encodings) convey the same information?  If the information is
different, what are the differences?    What would have have to remove
from each file to make them say the same thing?

      blindfold report file1.nt file2.nt

seems like a good start in this direction, although it only works on
N-Triples files right now.  (There are some in the "samples" directory,
and you can use Dave Becket's rdfdump [1] to get N-Triples from
RDF/XML.)

      blindfold --help

will give you some more useful information.

What makes this non-trivial is the bNodes (local existential variables,
anonymous nodes) in RDF which need to be matched between the files with
some consistent mapping.  My guiding notion is that the difference in
information conveyed (entailed) between "There is someone with red hair
and black shoes" and "There is someone with red hair" is the "black
shoes" part, whether or not "someone" denotes the same thing to the
speakers of each sentence.

This work is related to the graph synchronization problem, but does not
completely address it.  To do that, I need to communicate additional
restrictions about the mapping between the bNodes in the two graphs.  I
think I know how to do that, but I haven't implemented it yet.

Please try to follow-up to only www-rdf-interest or www-rdf-logic,
depending on which community seems more relevant to you.

    -- sandro

[1] http://xsb.sourceforge.net/
[2] http://www.redland.opensource.ac.uk/raptor/
Received on Thursday, 4 October 2001 11:47:45 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:51:52 GMT