studioevoque / fuxi Goto Github PK

Not sure exactly what circumstance causes the rules to be empty but the error 
*seems* harmless enough, perhaps related to debugging output.

Can be reproduced with the curate tool, 

    curate  -r examples/tagging.n3 aiddata-china

attached patch fixes it with no apparent unwanted side-effects

Need to port use of 4Suite-XML to use Amara 2 instead (and eventually, 
[http://wiki.xml3k.org/Akara Akara] for the web infrastructure)

Currently, the use of owl:complementOf in an OWL theory and a conversion to a 
ruleset via the DLP 
method is the only way to introduce negated antecedent into a ruleset.  Support 
for RIF BLD XML 
syntax should address this eventually

What steps will reproduce the problem?

Run FuXi.Rete.BuiltinPredicates from command-line (triggering unit test 
harness):

resulting in:

TypeError: unbound method __init__() must be called with Uniterm instance as 
first argument (got ExternalFunction instance instead)

$ FuXi --safety=loose --strictness=defaultDerived --idb=owl:sameAs --method=bfp 
 --why="ASK { ex:subject1 owl:sameAs ex:subject2 }" --debug  
--ns=ex=http://www.w3.org/2002/03owlt/InverseFunctionalProperty/premises001#  
--pDSemantics 
--builtinTemplates=http://fuxi.googlecode.com/hg/RuleBuiltinSPARQLTemplates.n3 
--dlp http://www.w3.org/2002/03owlt/InverseFunctionalProperty/premises001.rdf

Query triggered for  rdfs:range_query(?P ?C) :- bfp:evaluate(rule:3 0)
Failed query SELECT ?C {     a  <http://www.w3.org/2000/01/rdf-schema#range> ?C }
..snip...
  File "/Library/Python/2.5/site-packages/pyparsing-1.4.11-py2.5.egg/pyparsing.py", line 2297, in parseImpl
    raise maxException
pyparsing.ParseException: Expected "}" (at char 17), (line:1, col:18)

Installation documentation on the wiki is dated and incomplete, including stale 
links and references to the FuXi version in the old svn repository.  

Note: edits are underway.

vasiliy@midgard:~/devel/ld/fuxi/fuxi/test$ PYTHONPATH=../ python 
additionalDLPTests.py 
Traceback (most recent call last):
  File "additionalDLPTests.py", line 6, in <module>
    from FuXi.Rete.RuleStore import N3RuleStore,SetupRuleStore
  File "/usr/lib/python2.5/site-packages/PIL/__init__.py", line 1, in <module>
    #
  File "FuXi/Rete/Network.py", line 27, in <module>
  File "/home/vasiliy/devel/ld/fuxi/fuxi/FuXi/DLP/ConditionalAxioms.py", line 67
SyntaxError: Non-ASCII character '\xc2' in file 
/home/vasiliy/devel/ld/fuxi/fuxi/FuXi/DLP/ConditionalAxioms.py on line 67, but 
no encoding declared; see http://www.python.org/peps/pep-0263.html for details


Attached is a patch that fixes this for me.

What steps will reproduce the problem?
1. Test data as previously supplied seems to work top down but bottoms up 
results in exceptions for these two queries:

FuXi --input-format="n3" --ns=test=http://xdors.net/demo# 
--ns=owl=http://www.w3.org/2002/07/owl# --dlp \
    --method=bottomUp   \
    --why="SELECT ?x WHERE { ?x a test:female  }" scratch.n3
echo mensch ==========================================================
FuXi --input-format="n3" --ns=test=http://xdors.net/demo# 
--ns=owl=http://www.w3.org/2002/07/owl# --dlp \
    --method=bottomUp   \
    --why="SELECT ?x WHERE { ?x a test:mensch  }" scratch.n3


What is the expected output? What do you see instead?

Exceptions of the following form arise.  This seems suspiciously like the 
others I reported recently with keyErrors in convertTerm...


/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/Fu
Xi-1.0.dev-py2.6.egg/FuXi/Rete/AlphaNode.py:6: DeprecationWarning: the sets 
module is deprecated
  from sets import Set
/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/Fu
Xi-1.0.dev-py2.6.egg/FuXi/Rete/TopDown.py:12: DeprecationWarning: the md5 
module is deprecated; use hashlib instead
  import itertools, copy, md5
Time to build production rule (RDFLib): 7.70092010498e-05 seconds
Magic seed fact (used in bottom-up evaluation) :mensch_magic(?x)
/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/Fu
Xi-1.0.dev-py2.6.egg/FuXi/Rete/Magic.py:547: UserWarning: predicate symbol of 
test:has_child(?X ?OEucOySk25) is in both IDB and EDB. Marking as base
  "predicate symbol of %s is in both IDB and EDB. Marking as %s"%(term,mark))
/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/Fu
Xi-1.0.dev-py2.6.egg/FuXi/Rete/Magic.py:547: UserWarning: predicate symbol of 
test:person(?OEucOySk39) is in both IDB and EDB. Marking as base
  "predicate symbol of %s is in both IDB and EDB. Marking as %s"%(term,mark))
Traceback (most recent call last):
  File "/Library/Frameworks/Python.framework/Versions/2.6/bin/FuXi", line 8, in <module>
    load_entry_point('FuXi==1.0.dev', 'console_scripts', 'FuXi')()
  File "/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/FuXi-1.0.dev-py2.6.egg/FuXi/Rete/CommandLine.py", line 495, in main
    noMagic=noMagic):
  File "/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/FuXi-1.0.dev-py2.6.egg/FuXi/Rete/Magic.py", line 179, in MagicSetTransformation
    newRules.extend(AdditionalRules(factGraph))
  File "/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/FuXi-1.0.dev-py2.6.egg/FuXi/DLP/ConditionalAxioms.py", line 66, in AdditionalRules
    if tBox.query(FUNCTIONAL_PROPERTIES).askAnswer[0]:
  File "build/bdist.macosx-10.3-fat/egg/rdflib/Graph.py", line 738, in query
  File "build/bdist.macosx-10.3-fat/egg/rdflib/sparql/bison/Processor.py", line 50, in query
  File "build/bdist.macosx-10.3-fat/egg/rdflib/sparql/Algebra.py", line 312, in TopEvaluate
  File "build/bdist.macosx-10.3-fat/egg/rdflib/sparql/Algebra.py", line 165, in ReduceToAlgebra
  File "build/bdist.macosx-10.3-fat/egg/rdflib/sparql/bison/SPARQLEvaluate.py", line 273, in createSPARQLPConstraint
  File "build/bdist.macosx-10.3-fat/egg/rdflib/sparql/bison/SPARQLEvaluate.py", line 186, in mapToOperator
  File "build/bdist.macosx-10.3-fat/egg/rdflib/sparql/bison/SPARQLEvaluate.py", line 218, in mapToOperator
  File "build/bdist.macosx-10.3-fat/egg/rdflib/sparql/bison/SPARQLEvaluate.py", line 71, in convertTerm
KeyError: u'owl'

The current BFP reasoner does not combine atomic inclusion axioms that share 
the same head as the SLD reasoner does.  As a result, for the same test case 
below, whereas the SLD reasoner dispatches 1 query, the BFP reasoner dispatches 
2:

$ python testOWL.py --strategy=sld --debug --singleTest=OWL/unionOf/premises001
..snip..

## Full SPARQL Algebra expression ##
BGP((?SUBJECT,rdf:type,first:A))
###################################
    Solving ns1:A(?SUBJECT) {}
    Combining atomic inclusion axioms: 
[Forall ?X ( ns1:A_f(?X) :- first:Human(?X) ),
 Forall ?X ( ns1:A_f(?X) :- first:Animal(?X) )]
SELECT ?SUBJECT {
  ?SUBJECT a ?KIND
  FILTER(
    ?KIND = first:Human ||
    ?KIND = first:Animal)
}
{'OWL/unionOf/Manifest001.rdf': '59.7569942474 milli seconds'}

$ python testOWL.py --strategy=bfp --debug --singleTest=OWL/unionOf/premises001
..snip..
<Network: 5 rules, 6 nodes, 6 tokens in working memory, 1 inferred tokens> 
<TerminalNode with 1 actions (ns1:A_f(?X) :- And( ns1:OpenQuery(ns2:A) 
bfp:evaluate(rule:1 1) )) : CommonVariables: [] (1 in left, 1 in right 
memories)>
    ns1:A_f(?X) :- And( ns1:OpenQuery(ns2:A) bfp:evaluate(rule:1 1) )
        1 instanciations

@prefix ns1: <http://www.w3.org/2002/03owlt/unionOf/premises001#>.

 ns1:John a ns1:A. 
SELECT ?X {     ?X  a  ns1Animal }
SELECT ?X {     ?X  a  ns1:Human }
{'OWL/unionOf/Manifest001.rdf': '327.481985092 milli seconds'}

Rules with second order predicates of the form ?C(:Foo :Bar) are not supported 
by tip-down or GMS algorithms

[http://code.google.com/p/fuxi/wiki/ReteActions Externally defined actions] are 
associated with terminal nodes but there is no easy way to get the terminal 
node associated with a particular rule.  The network instances have a 
terminalNodes instance that is a list of terminal nodes but probably could use 
a dictionary that maps rules (used to build the network) to the corresponding 
terminal node in order to make it easier for the user to register externally 
defined actions.

What steps will reproduce the problem?
1. Use layercake-python with FuXi trunk

    python testOWL.py --strategy=sld 

What is the expected output? What do you see instead?

{{{
  File "/Library/Python/2.5/site-packages/FuXi-1.1.production-py2.5.egg/FuXi/Rete/TopDown.py", line 480, in collectAnswers
    for ans,ns in _ans:
  File "/Library/Python/2.5/site-packages/FuXi-1.1.production-py2.5.egg/FuXi/Rete/TopDown.py", line 583, in SipStrategy
    for answers in memoizeMemory[queryLiteral]:
RuntimeError: Set changed size during iteration
----------------------------------------------------------------------
Ran 1 test in 2.047s
}}}

Please use labels and text to provide additional information.

Using the following command and input I get successful results for
some queries but not others.  Can you advise how to resolve, debug or
work around such issues?  Using the same command line except varying
the queries as follows

1) SELECT ?x WHERE { ?x a test:party    }
2) SELECT ?x WHERE { ?x a test:person }
3) SELECT ?x WHERE { ?x a test:female  }
4) SELECT ?x WHERE { ?x a test:mensch }

FuXi-1.0.dev returns results as expected for (1) and (2); but for
(3) and (4) error messages of the following form are returned:

AssertionError: [(?x, rdflib.URIRef('http://www.w3.org/1999/02/22-rdf-
syntax-ns#type'), rdflib.URIRef('http://xdors.net/demo#mensch'))]

Current implementation isn't moving all builtins to the end

Diff of Network.py that *should* fix problem - but introduces performance drop

{{{
diff -r 4f1111ec2d4d lib/Rete/Network.py
--- a/lib/Rete/Network.py       Fri Oct 02 15:11:36 2009 -0400
+++ b/lib/Rete/Network.py       Thu Oct 29 22:57:47 2009 -0400
@@ -657,10 +658,11 @@
                     self._resetinstanciationStats()                        
                 #self.checkDuplicateRules()
                 return
+            isaBuiltin = self.ruleStore.filters.get(currentPattern[PREDICATE])
             if HashablePatternList([currentPattern]) in self.nodes:
                 #Current pattern matches an existing alpha node
                 matchedPatterns.append(currentPattern)
-            elif matchedPatterns in self.nodes:
+            elif not isaBuiltin and matchedPatterns in self.nodes:
                 #preceding patterns match an existing join/beta node
                 newNode = self.createAlphaNode(currentPattern)
                 if len(matchedPatterns) == 1 and HashablePatternList([None])+matchedPatterns in 
self.nodes:

}}}

What steps will reproduce the problem?
1. Fix for empty dictionary co ntains garbage characters so for example 
testExistentialInHead.py won't ru:  Lexical error at line 67, column 55.  
Encountered: "\u00ac" (172), after : "" ConditionalAxioms.py    /fuxi/lib/DLP   line 
66

What steps will reproduce the problem?
1. testOwl  --strategy gms

What is the expected output? What do you see instead?

======================================================================
ERROR: testOwl (__main__.OwlTestSuite)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/Users/Shared/workspaces/workspace/fuxi/test/testOWL.py", line 361, in testOwl
    conclusionFile)
  File "/Users/Shared/workspaces/workspace/fuxi/test/testOWL.py", line 248, in MagicOWLProof
    timing=self.calculateEntailments(factGraph)
  File "/Users/Shared/workspaces/workspace/fuxi/test/testOWL.py", line 152, in calculateEntailments
    self.network.feedFactsToAdd(generateTokenSet(factGraph))
  File "/Users/Shared/workspaces/workspace/fuxi/build/lib/FuXi/Rete/Network.py", line 604, in feedFactsToAdd
    self.addWME(token)
  File "/Users/Shared/workspaces/workspace/fuxi/build/lib/FuXi/Rete/Network.py", line 594, in addWME
    alphaNode.activate(wme.unboundCopy())
  File "/Users/Shared/workspaces/workspace/fuxi/build/lib/FuXi/Rete/AlphaNode.py", line 322, in activate
    memory.successor.propagate(memory.position,aReteToken.debug,wme=aReteToken)
  File "/Users/Shared/workspaces/workspace/fuxi/build/lib/FuXi/Rete/BetaNode.py", line 660, in propagate
    self._activate(PartialInstanciation([wme],consistentBindings=wme.bindingDict.copy()),debug)
  File "/Users/Shared/workspaces/workspace/fuxi/build/lib/FuXi/Rete/BetaNode.py", line 607, in _activate
    self.network.fireConsequent(partInstOrList,self,debug)
  File "/Users/Shared/workspaces/workspace/fuxi/build/lib/FuXi/Rete/Network.py", line 553, in fireConsequent
    self.addWME(inferredToken)
  File "/Users/Shared/workspaces/workspace/fuxi/build/lib/FuXi/Rete/Network.py", line 594, in addWME
    alphaNode.activate(wme.unboundCopy())
  File "/Users/Shared/workspaces/workspace/fuxi/build/lib/FuXi/Rete/AlphaNode.py", line 322, in activate
    memory.successor.propagate(memory.position,aReteToken.debug,wme=aReteToken)
  File "/Users/Shared/workspaces/workspace/fuxi/build/lib/FuXi/Rete/BetaNode.py", line 652, in propagate
    print self.clauseRepresentation()
  File "/Users/Shared/workspaces/workspace/fuxi/build/lib/FuXi/Rete/BetaNode.py", line 557, in clauseRepresentation
    return repr(first(self.rules).formula)
AttributeError: 'NoneType' object has no attribute 'formula'

----------------------------------------------------------------------
Ran 1 test in 0.798s

FAILED (errors=1)

FuXi used to produce PML OWL/RDF documents using this function 
(FuXi.Rete.Proof.GenerateProof) module:

http://code.google.com/p/fuxi/source/browse/lib/Rete/Proof.py#151

This has not been tested since the addition of the BFP query answering method 
was added.  

Alternatively, callbacks can be registered as actions 
(http://code.google.com/p/fuxi/wiki/ReteActions) that keep track of which rule 
fired which ground fact.

Purpose of code changes on this branch:


When reviewing my code changes, please focus on:

is it well written, do examples work etc.

After the review, I'll merge this branch into:
/trunk

Currently, most of the inference heuristics do not support second order 
predicates (predicates symbols that are variables in RIF frames / positional 
terms: ?C(:Foo :Bar  )   ).  These, however in theory can be supported via 
specialization.  Consider the following from the third owl:oneOf OWL 
[http://www.w3.org/TR/owl-test/byFunction#function-oneOf test case]:

Source graph:

{{{
@prefix first: <http://www.w3.org/2002/03owlt/oneOf/premises003#>.
@prefix owl: <http://www.w3.org/2002/07/owl#>.
@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>.

 first:T2 owl:oneOf ( first:large first:medium first:small ). 

 first:myT a first:T1. 

 first:T1 owl:oneOf ( first:small first:medium first:large ). 
}}}

The following rule should apply to the query for first:T2(first:MyT):

{{{
Forall ?X ?C ?L ( ?C(?X) :- And( owl:oneOf(?C ?L) list:in(?X ?L) ) )
}}} 

In particular, for every statement in the OWL/RDF graph where the predicate is 
owl:oneOf, a specific rule can be generated for the subject (first:T2 in this 
case):

{{{
Forall ?X ?C ?L ( first:T2(?X) :- And( owl:oneOf(first:T2  ?L) list:in(?X ?L) ) 
)
}}}

Currently the answers from queries dispatched by the backward chainer are not 
being iterated over 
lazily.  

What steps will reproduce the problem?

testOwl.py --strategy=bfp

The test gets underway then hangs in what appears to be an infinite loop, 
suckng up 100% of the CPU.  However, it finally completes after a few minutes.  
This is of course much more cpu intensive than with the other strategies:


===============================================
 'OWL/intersectionOf/Manifest001.rdf': '333.187818527 milli seconds',
 'OWL/inverseOf/Manifest001.rdf': '72.2098350525 milli seconds',
 'OWL/unionOf/Manifest001.rdf': '67.4600601196 milli seconds'}
ok

----------------------------------------------------------------------
Ran 1 test in 205.390s

OK
===============================================


Here's what the output looks like when it hangs:


===============================================
 Forall ?S ?O ( owl:complementOf(?O ?S) :- owl:complementOf(?S ?O) ),
 Forall ?S ?O ( owl:disjointWith(?O ?S) :- owl:disjointWith(?S ?O) ),
 Forall ?L1 ?L2 ?ANY ?L ( owl:differentFrom(?L1 ?L2) :- And( owl:AllDifferent(?ANY) owl:distinctMembers(?ANY ?L) list:in(?L1 ?L) list:in(?L2 ?L) <http://www.w3.org/2000/10/swap/log#notEqualTo>(?L1,?L2) ) ),
 Forall ?I ?R ?L ( list:in(?I ?L) :- And( rdf:rest(?L ?R) list:in(?I ?R) ) ),
 Forall ?I ?L ( list:in(?I ?L) :- rdf:first(?L ?I) )]
Goal to solve  SELECT ?SUBJECT {    ?SUBJECT 
<http://www.w3.org/2002/07/owl#differentFrom> 
<http://www.w3.org/2002/03owlt/distinctMembers/premises001#Barney> }
Time to build production rule (RDFLib): 8.89301300049e-05 seconds
===============================================


This is entered with an initial priority of high since bfp is now the preferred 
reasoning strategy.

What steps will reproduce the problem?
1. Prepare a vocabulary with an rdfs:subPropertyOf relation. An example is 
attached as subprop-tbox.n3.
2. Prepare instance data that uses a subproperty from that vocabulary. An 
example is attached as subprop-abox.n3.
3. Use FuXi to materialize inferences from this data. An example program for 
this is attached as reason.py (`python reason.py subprop-tbox.n3 
subprop-abox.n3`).

What is the expected output? What do you see instead?
I expect to see the inferred superproperty statement, among other RDFS 
entailments. I do not see such a statement.

What version of the product are you using? On what operating system?
cset a2e7663d59cd

Please provide any additional information below.
Note that other RDFS entailments (superclass, domain, range) are performed as 
they should.
I observe this also on "real" vocabularies like DC (e.g. dcterms:title is a 
subproperty of dc:title).

The BFP query mediation can sometimes result in duplicating mediated queries.  
A simple memoization mechanism is needed to eliminate this redundancy so 
queries that are equivalent up to renaming of variables are only evaluated once 
against the base RDF dataset

Traceback (most recent call last):
  File ".../layercake-python/rdflib/sparql/Algebra.py", line 1273, in testSimpleGraphPattern
    from rdflib.sparql.bison.Processor import Parse
ImportError: cannot import name Parse

Usually systems that provide some form of closed world assumption (or 
non-monotonic) negation 
have a way to ensure the program (the rules and facts) are stratified (i.e., 
there are no circular 
references between antecedents and consequents of rules).  This guarantees 
unique, consistent 
answers for queries involving default negation

Need documentation specific to the use of hybrid predicates and their effect on 
the reasoning process

What steps will reproduce the problem?
1. 2 rdflib graphs: first empty (g1), the second (g2) has Restriction (r)
on a DatatypeProperty
2. r.serialize(g1)
3. in g1 the datatypeProperty is also ObjectProperty (the creation of the
copy of the property does not set the type of property)

What is the expected output? What do you see instead?
g1 has a copy of the property that is ONLY datatypeProperty

What version of the product are you using? On what operating system?
latest

Please provide any additional information below.
correct the serialize method in the Restriction class to copy type of
property (default is objectProperty)

What steps will reproduce the problem?
1. comment out rdflib OWL import
2. execute testOWL.py

What is the expected output? What do you see instead?

testOWL.py:2: DeprecationWarning: the sets module is deprecated
  from sets import Set
/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/Fu
Xi-1.0.dev-py2.6.egg/FuXi/Rete/SidewaysInformationPassing.py:7: 
DeprecationWarning: the md5 module is deprecated; use hashlib instead
  import unittest, os, sys, itertools, md5
testOwl (__main__.OwlTestSuite) ... Time to build production rule (RDFLib): 
0.000104188919067 seconds
OWL/AllDifferent/premises001
OWL/AllDifferent/conclusions001
<OWL/AllDifferent/conclusions001.rdf> :- <OWL/AllDifferent/premises001.rdf>
OWL/AllDifferent/premises001 http://www.w3.org/2002/07/owl#AllDifferent 
using <code xmlns="http://www.w3.org/1999/xhtml">AllDifferent</code> to derive 
<code xmlns="http://www.w3.org/1999/xhtml">differentFrom</code>
Original program
OWL/AllDifferent/Manifest001.rdf OWL/AllDifferent/premises001
feature:  http://www.w3.org/2002/07/owl#AllDifferent

using <code xmlns="http://www.w3.org/1999/xhtml">AllDifferent</code> to derive 
<code xmlns="http://www.w3.org/1999/xhtml">differentFrom</code>
[]
ERROR

======================================================================
ERROR: testOwl (__main__.OwlTestSuite)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "testOWL.py", line 327, in testOwl
    timings.append(self.MagicOWLProof(goals,program,factGraph,conclusionFile))
  File "testOWL.py", line 164, in MagicOWLProof
    for rule in AdditionalRules(factGraph):
  File "/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/FuXi-1.0.dev-py2.6.egg/FuXi/DLP/ConditionalAxioms.py", line 66, in AdditionalRules
    if tBox.query(FUNCTIONAL_PROPERTIES).askAnswer[0]:
  File "build/bdist.macosx-10.3-fat/egg/rdflib/Graph.py", line 738, in query
    extensionFunctions))
  File "build/bdist.macosx-10.3-fat/egg/rdflib/sparql/bison/Processor.py", line 50, in query
    extensionFunctions=extensionFunctions)
  File "build/bdist.macosx-10.3-fat/egg/rdflib/sparql/Algebra.py", line 312, in TopEvaluate
    None)
  File "build/bdist.macosx-10.3-fat/egg/rdflib/sparql/Algebra.py", line 165, in ReduceToAlgebra
    prolog))
  File "build/bdist.macosx-10.3-fat/egg/rdflib/sparql/bison/SPARQLEvaluate.py", line 273, in createSPARQLPConstraint
    for expr in reducedFilter]))
  File "build/bdist.macosx-10.3-fat/egg/rdflib/sparql/bison/SPARQLEvaluate.py", line 186, in mapToOperator
    mapToOperator(expr.right,prolog,combinationArg,constraint=constraint),
  File "build/bdist.macosx-10.3-fat/egg/rdflib/sparql/bison/SPARQLEvaluate.py", line 218, in mapToOperator
    return "'%s'"%convertTerm(expr,prolog)
  File "build/bdist.macosx-10.3-fat/egg/rdflib/sparql/bison/SPARQLEvaluate.py", line 71, in convertTerm
    return URIRef(queryProlog.prefixBindings[term.prefix] + term.localname)
KeyError: u'owl'

----------------------------------------------------------------------
Ran 1 test in 0.258s

What steps will reproduce the problem?
1. Prepare a set of instance data with an `:a owl:sameAs :b` relation, but no 
`:b owl:sameAs :a` relation. An example is attached as sameas.n3.
2. Use FuXi's setupDescriptionLogicProgramming() to intialize a network on any 
set of TBox data, for example on an empty graph.
3. Materialize inferences from the instance data.

What is the expected output? What do you see instead?
I expect all statements about :a to be "copied" to :b, and vice-versa. Instead, 
I see that the statements about :b are not "copied" to :a.

What version of the product are you using? On what operating system?
cset 781be253fe4b

Please provide any additional information below.
The attached patch (sameas.diff) fixes this for me by introducing an explicit 
rule into `non_DHL_OWL_Semantics` that makes owl:sameAs symmetric. I don't know 
if this is the right way to handle it.

The diff says it all:

diff -r ade3a47cc712 lib/Syntax/InfixOWL.py
--- a/lib/Syntax/InfixOWL.py    Mon Jun 07 20:53:49 2010 -0400
+++ b/lib/Syntax/InfixOWL.py    Tue Jul 20 14:18:30 2010 +0100
@@ -769,9 +769,9 @@
         Chaining 3 intersections

         >>> female      = Class(exNs.Female,graph=g)
-        >>> male        = Class(exNs.Human,graph=g)
+        >>> human       = Class(exNs.Human,graph=g)
         >>> youngPerson = Class(exNs.YoungPerson,graph=g)
-        >>> youngWoman = female & male & youngPerson
+        >>> youngWoman = female & human & youngPerson
         >>> youngWoman
         ex:YoungPerson that ( ex:Female and ex:Human )
         >>> isinstance(youngWoman,BooleanClass)

FuXi cannot support psuedo-2nd order predicate SPARQL entailment queries such 
as:

SELECT * { ex:Foo ?predicate ex:Bar } 

or 

SELECT * { ex:Foo a ?predicate } 

Primarily, the BFP method needs a principled solution to implement

See: 
http://answers.semanticweb.com/questions/1594/what-are-the-semantics-of-2nd-orde
r-predicates-in-rdf-rule-languages

Currently the backward and forward inference procedure only supports 
user-specified filters, i.e., 
there are two inputs and the result is either true or false based on the 
user-supplied computation.  
We want to be able to add support for user-specified computations that result 
in arbitrary RDF 
terms. 

if newInferredFacts is the empty graph, the test will fail and
self.inferredFacts will be initialised to a fresh instance

diff -r ade3a47cc712 lib/Rete/Network.py
--- a/lib/Rete/Network.py       Mon Jun 07 20:53:49 2010 -0400
+++ b/lib/Rete/Network.py       Thu Jul 22 17:38:55 2010 +0100
@@ -442,7 +442,7 @@
                 node.memories[RIGHT_MEMORY].reset()
         self.justifications = {}
         self.proofTracers = {}
-        self.inferredFacts = newinferredFacts and newinferredFacts or Graph()
+        self.inferredFacts = newinferredFacts is not None and newinferredFacts 
or Graph()
         self.workingMemory = set()
         self._resetinstanciationStats()        

$ FuXi --safety=loose --strictness=defaultDerived \
--idb=owl:sameAs --method=both  \
--why="ASK { ex:subject1 owl:sameAs ex:subject2 }" \
--debug  \
--ns=ex=http://www.w3.org/2002/03owlt/InverseFunctionalProperty/premises001#  \
--pDSemantics 
--builtinTemplates=http://fuxi.googlecode.com/hg/RuleBuiltinSPARQLTemplates.n3 \
--dlp http://www.w3.org/2002/03owlt/InverseFunctionalProperty/premises001.rdf

SELECT ?P {     ?P owl:inverseOf owl:sameAs_query_bb } 1 apriori binding(s)-> []
Query triggered for  rdfs:range_query(?P ?C) :- bfp:evaluate(rule:3 0)
Traceback (most recent call last):
  File "/Library/Python/2.5/site-packages/FuXi-1.1.production-py2.5.egg/FuXi/LP/BackwardFixpointProcedure.py", line 207, in __call__
    queryStr,rt = edbQuery.evaluate(self.bfp.debug)
    raise maxException
pyparsing.ParseException: Expected "}" (at char 17), (line:1, col:18)

Patch attached. Only test failing is the mysterious PROGRAM2

Switch FuXi to using the latest version of rdflib that includes changes to the 
module structure, etc..

What steps will reproduce the problem?
1.
2.
3.

What is the expected output? What do you see instead?

/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/Fu
Xi-1.0.dev-py2.6.egg/FuXi/Rete/AlphaNode.py:6: DeprecationWarning: the sets 
module is deprecated
  from sets import Set
Time to build production rule (RDFLib): 9.89437103271e-05 seconds
/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/Fu
Xi-1.0.dev-py2.6.egg/FuXi/Rete/SidewaysInformationPassing.py:7: 
DeprecationWarning: the md5 module is deprecated; use hashlib instead
  import unittest, os, sys, itertools, md5
E
======================================================================
ERROR: testUnionSkolemization (__main__.UnionSkolemizedTest)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "testSkolemization.py", line 31, in testUnionSkolemization
    p=network.setupDescriptionLogicProgramming(self.tBoxGraph)
  File "/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/FuXi-1.0.dev-py2.6.egg/FuXi/Rete/Network.py", line 352, in setupDescriptionLogicProgramming
    for rule in AdditionalRules(owlN3Graph):
  File "/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/FuXi-1.0.dev-py2.6.egg/FuXi/DLP/ConditionalAxioms.py", line 66, in AdditionalRules
    if tBox.query(FUNCTIONAL_PROPERTIES).askAnswer[0]:
  File "build/bdist.macosx-10.3-fat/egg/rdflib/Graph.py", line 738, in query
    extensionFunctions))
  File "build/bdist.macosx-10.3-fat/egg/rdflib/sparql/bison/Processor.py", line 50, in query
    extensionFunctions=extensionFunctions)
  File "build/bdist.macosx-10.3-fat/egg/rdflib/sparql/Algebra.py", line 312, in TopEvaluate
    None)
  File "build/bdist.macosx-10.3-fat/egg/rdflib/sparql/Algebra.py", line 165, in ReduceToAlgebra
    prolog))
  File "build/bdist.macosx-10.3-fat/egg/rdflib/sparql/bison/SPARQLEvaluate.py", line 273, in createSPARQLPConstraint
    for expr in reducedFilter]))
  File "build/bdist.macosx-10.3-fat/egg/rdflib/sparql/bison/SPARQLEvaluate.py", line 186, in mapToOperator
    mapToOperator(expr.right,prolog,combinationArg,constraint=constraint),
  File "build/bdist.macosx-10.3-fat/egg/rdflib/sparql/bison/SPARQLEvaluate.py", line 218, in mapToOperator
    return "'%s'"%convertTerm(expr,prolog)
  File "build/bdist.macosx-10.3-fat/egg/rdflib/sparql/bison/SPARQLEvaluate.py", line 71, in convertTerm
    return URIRef(queryProlog.prefixBindings[term.prefix] + term.localname)
KeyError: u'owl'

----------------------------------------------------------------------
Ran 1 test in 0.170s



Please use labels and text to provide additional information.

What steps will reproduce the problem?
1. Import error occurs with clean installation of FuXi on 7-1-10

This appears to be an extraneous import statement that can  be commented out.  
The additional errors I will report are net of this workaround.

it would be nice (and straightforward) to be able to do:

$ FuXi --input-format=n3 --output=rif --rules=transitive-test.n3 Time to build 
production rule Forall ?Y ?X ?Z ( ex:hasAncestor(?X ?Z) :- And( 
ex:hasAncestor(?X ?Y) ex:hasAncestor(?X ?Z) ) )


$ FuXi --input-format=n3 --output=spin --rules=transitive-test.n3 Time to build 
production 

@prefix owl: <http://www.w3.org/2002/07/owl#> .
@prefix ex: <http://example.com/terms#> .
@prefix sp: <http://spinrdf.org/sp>
@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>.

[ a  sp:Construct ;
      sp:templates ([ sp:object sp:_Z ;
                      sp:predicate ex:hasAncestor ;
                      sp:subject spin:_X
                    ]) ;
      sp:where ([ sp:object spin:_Y ;
                  sp:predicate ex:hasAncestor ;
                  sp:subject sp:_X
                ] 
                [ sp:object sp:_Z ;
                  sp:predicate ex:hasAncestor ;
                  sp:subject sp:_Y
                ])
    ]

Where transitive-test.n3 is:

@prefix owl: <http://www.w3.org/2002/07/owl#> .
@prefix ex: <http://example.com/terms#> .

{ ?X ex:hasAncestor ?Y . ?X ex:hasAncestor ?Z } => { ?X ex:hasAncestor ?Z } .

This way n3 rules can be 'vizualized' within TopBraid composer, The OWL2 RL in 
RIF templateRules  algorithm (which FuXi supports) can be leveraged to generate 
efficient, custom SPIN rules from an OWL 2 RL ontology

What steps will reproduce the problem?
1.  easy_install-2.5 fuxi

What is the expected output? What do you see instead?

When attempting easy_install on OS X 10.6.6, does not run clean but produces:
...
Processing FuXi-1.1.production.macosx-10.5-i386.tar.gz
error: Couldn't find a setup script in 
/var/folders/yN/yNVCI5CMGGm7dmdKQkYp-++++TI/-Tmp-/easy_install-4KltzK/FuXi-1.1.p
roduction.macosx-10.5-i386.tar.gz

Workaround is to install manually from sources as documented on the wiki, see
http://code.google.com/p/fuxi/wiki/Installation_Testing.

From Uche: I've been going through all my projects and eliminating the need for 
setuptools.  I'd recommend the same for FuXi. 

Adding support to FuXi for RIF XML syntax

In particular if we have Exists then it breaks because it can't be transformed 
into a triple.

Patch attached.

In order to fully deprecate the (currently faulty) SLD / topdown reasoner, the 
new BFP reasoner needs to support all its capabilities.

The attached is intended as a first step toward in providing set of FuXi demos. 
 The thought is to provide command line demonstrations of simple queries 
against OWL-based ontologies and facts.  These will later be supplemented with 
demo programs, rule examples, etc.

While the attached inputs produce expected results under third party products, 
results vary under FuXi's different strategies.  

The file attached file, output.txt, reflects the results of the shell script 
Fuxi-bsa.command.   The queries ask who is a man and who is a person given a 
minimal set of facts, i.e., john is a man and jane is a woman.   The classes 
man and woman are subclasses of adult which in turn is a subclass of person--so 
we can infer john and jane are persons.  

Results under various strategies are as follows:
* SLD yields the expected answers to both queries.
* GMS does not report results in either case.
* BFP reports the expected result for who is a man; but aborts with the query 
about who is a person with the exception "TypeError: argument of type 
'NoneType' is not iterable."  

Perhaps the TypeError issue is the same BFP problem as alluded to in the last 
paragraph of Issue 44.

This is entered initially as a single, new, low priority report since the 
issues did not arise in a production environment and also because the single 
set of inputs provide grounds for comparison across strategies.  However there 
appear to be two distinct issues here  (1) missing results for GMS queries and 
(2) a NoneType is not iterable issue, which while mentioned in issue 44 is not 
the main concern of that report.

What steps will reproduce the problem?
1. click on the links at bottom of page
2.
3.

What is the expected output? What do you see instead?
Since site is about a the semantic web, I expected to see html

What version of the product are you using? On what operating system?
chrome 6 osX

Please provide any additional information below.

Just a heads up.  

FuXi's SPARQL query mediation engine needs further testing using the recently 
created SPARQL 1.1 entailment regime tests:

http://www.w3.org/2009/sparql/docs/tests/data-sparql11/entailment/

A harness similar to the one created for the OWL tests needs to be created

This will be done over the next couple of days

Hi FuXi creators,
first of all, don't worry, this is not an issue. I just wanted to tell you 
about my "bundle". I've been adapting rdflib 3.1 with rdfextras, FuXi and the 
store modules of rdflib 2.4. The result is RDF.py: 
https://github.com/fertapric/RDF.py.

In addition, I added some examples.

your thoughts about the project? names/licenses of the project (to update the 
correspondent files and credits)??
Best regards


-- Fernando Tapia

Running a variant of the example from  
http://code.google.com/p/fuxi/wiki/FuXiUserManual against hg 9ef1f7f348b5 using 
layercake-python SVN r354,

> python ./lib/Rete/CommandLine.py --debug \
>   --ruleFacts --dlp --output=conflict --method=sld --strict=defaultDerived \
>   --ns='test=http://www.w3.org/2002/03owlt/TransitiveProperty/premises001#' \
>   --why="ASK { test:Ghent test:path <http://www.example.com/nopathhere> }" \
>   http://www.w3.org/2002/03owlt/TransitiveProperty/premises00

(the sha1sum of http://www.w3.org/2002/03owlt/TransitiveProperty/premises00 as 
retrieved is 918a841b1404ba797848717e939a068c5cfe6fa6) produces

> Time to reach answer ground goal answer of True: 217.927932739 milli seconds

Which is clearly wrong: there isn't a test:path from test:Ghent to that URI.  
Changing the subject to test:Antwerp also produces True, but test:Amsterdam 
produces False.

Using --method=gms I get output (prior to the profiling data) ending with

> 
> Time to calculate closure on working memory:  2.85601615906 milli seconds
> <TerminalNode (ns1:path_magic(?X) :- ns1:path_magic(?X ?IBYGKzCx20)) (pass-
> thru): CommonVariables: [?X, ?IBYGKzCx20] (0 in left, 1 in right memories)>
>   ns1:path_magic(?X) :- ns1:path_magic(?X ?IBYGKzCx20)
>       1 instanciations
>         485036 function calls (260105 primitive calls) in 6.895 CPU seconds

in every case, both those which should produce a path and which shouldn't.

Using --method=bfp throws "TypeError: argument of type 'NoneType' is not 
iterable", and --method=naive (with --output=conflict and --output=n3) seems to 
produce no meaningful output whatsoever.

What steps will reproduce the problem?
1. Fix for empty dictionary co ntains garbage characters so for example 
testExistentialInHead.py won't ru:  Lexical error at line 67, column 55.  
Encountered: "\u00ac" (172), after : "" ConditionalAxioms.py    /fuxi/lib/DLP   line 
66