Blindfold

Status

This is experimental work. Some of it may be useful to people working with RDF, DAML, and other Semantic Web technologies. Discussion of this is welcome on www-rdf-interest@w3.org.

Update

After some months of lying fallow, it's become clear to me that

1. blindfold is just blindfold grammars; the current bf code base is really blindfold plus a C++ RDF library. The "diff" functionality is really a different app which happens to be built with both libraries.
2. the semantic annotation language for blindfold grammars should be arbitrary, with a meta-syntax interface, so that blindfold is just an easy way to make efficient language translators. It translates well between tree-structured languages, but doesn't address re-ordering (which is fine for RDF languages).

Some rough notes to myself on this:

command line:

 blindfold (no args needed)
       -xml or -plaintext-tagged
       -auto (plaintext-tagged, then xml, then ...?)
	     internally-identified-language



*  @xmlns instead of prefix?
      @set xmlns:rdf="...."
*  @language(" ... ntriples ... ")
   or @set language="...."
   @set annotationLanguage="http://www.w3.org/2002/05/n-triples#spec";
   @set annotationLanguage="http://www.w3.org/2002/05/rdflang#ntriples";
   @set annotationLanguage="urn:uuid:458a4754-5fdf-11d6-9376-0001f4ec99c8";
        annLang=
	outLang
   ...
*  @local(foo)    same as  foo    where elsewhere  foo::= ;

   @buffer(strexpr)
   @clearBuffer
      $bufferContents
  or
   @appendTo(symbol, text-expr)
*  @strcat(symbol, text-expr)
       which makes $text(sym) be longer by text-expr (eval'd now)
   @set text(foo)=text(foo)+"hello";

*  $foo and $text(foo) inside @{ ... }
   \{ and \} to include a { or } inside there.   or double them?
*  $$, {{, }}           [maybe only if unbalanced]

   turns $foo into whatever you set symbolTemplate to be with
   the $counter() part substituted.

   @set(_symbolTemplate, "_:bnf_$counter()")
*  @set(_symbolTemplate, "



SOME OBSOLETE STUFF FOLLOW...



Overview


Blindfold is a semantic web software toolkit.  It can be used by
programmers as a library of handy functions or at the command line to
perform various data transformations.



I've tried hard to keep things are simple as possible, following
the basic simplicity of the RDF model.


I generally treat RDF graphs as logical formulas, with existential
variables for bNodes (anonymous nodes).  An RDF graph merge is logical
conjunction.  Graph equivalence is equality of entailment (with
existential instantiation being the non-trivial inference method).  A
graph "difference" (A-B) asks the question "what is the smallest
formula I could conjoin with B to make it entail A?"


Command Line

The blindfold command can be used to perform a variety of
general operations on knowledge encode as RDF graphs.   With no
parameters, it will give you general usage instructions which may be
enough to get you started.   In v0.1.1 all input and output is in
N-Triples format.

Programming Interface

Blindfold's functionality is available in a general C++ library.
The documentation, and general API architecture sometimes lag behind
the current operational model.  The on-line
documentation is generated out of comments in the source code.

Grammar System


One of blindfold's more interesting features is its grammar system, which aims to demonstrate the
practical equivalence of many languages.  This system is present in a
pre-alpha state in the 0.1.1 release, and is not available from the
command line yet.


Download

Several alternatives:

Version 0.1.1 source (gzip'd tar file.).
Version 0.1.1
binary (linux i386 executable; just "chmod o+x blindfold" and put
it somewhere in your path.))
Our cvs repository
 has the latest code, which doesn't always work.


You need XSB installed
(and in your path) for blindfold 0.1.1 to work properly.  


Future Directions



Give the user control over how bNodes are mapped between graphs,
to allow reproducable "patches" to be produced and applied.

Integrate and improve the grammar system, using it for RDF graph
I/O and for communicating with external agents (like XSB).

Logical inference, to do smart things with daml:equivaleTo,
daml:UniqueProperty and with Horn Logics (a la cwm), schema/ontology
validation.

Make this all faster, better tested, easier to install, not
reliant on XSB, maybe even ported beyond unix.

Support communication with RDF repositories other than files.  The
XSBAgent design is an early attempt in this direction; the idea is
that datasets (for input, output, or modification) might reside on
remote system and be handled without ever being transported here.



 

Sandro Hawke


$Date: 2002/05/21 12:53:07 $