def computeHybridOntology( ff, topConcepts ):
""" computes the hybrid ontology
@param[in] ff list of input ontologies
@param[in] topConcepts concepts which are required to participate in
every hybrid ontology relation
@returns a hybrid ontology which contains all relations found in the
ontologies ff between concepts listed in the topConcept list
"""
g = ConjunctiveGraph()
allTopConcepts = set( topConcepts )
usedTopConcepts = set()
for f in ff:
for s, p, o in extractSPO( XCL2RDF.toRDF(open(f).read() ) ):
if s in allTopConcepts and o in allTopConcepts:
g.add( (getUrl(s), NS_RDFS['label'], Literal(s)) )
g.add( (getUrl(p), NS_RDFS['label'], Literal(p)) )
g.add( (getUrl(o), NS_RDFS['label'], Literal(o)) )
g.add( (getUrl(s), getUrl(p), getUrl(o)) )
usedTopConcepts.add( s )
usedTopConcepts.add( o )
_addUseCaseSpecificUnusedConcepts(g)
with open("hybrid-graph.rdf", "w") as f:
f.write( g.serialize() )
unusedConcepts = allTopConcepts.difference( usedTopConcepts )
print "# of unused concepts: %d" % len( unusedConcepts )
print ", ".join( list(unusedConcepts) )
def computeHybridOntology( ff, topConcepts ):
""" computes the hybrid ontology
@param[in] ff list of input ontologies
@param[in] topConcepts concepts which are required to participate in
every hybrid ontology relation
@returns a hybrid ontology which contains all relations found in the
ontologies ff between concepts listed in the topConcept list
"""
g = ConjunctiveGraph()
allTopConcepts = set( topConcepts )
usedTopConcepts = set()
for f in ff:
for s, p, o in extractSPO( XCL2RDF.toRDF(open(f).read() ) ):
if s in allTopConcepts and o in allTopConcepts:
g.add( (getUrl(s), NS_RDFS['label'], Literal(s)) )
g.add( (getUrl(p), NS_RDFS['label'], Literal(p)) )
g.add( (getUrl(o), NS_RDFS['label'], Literal(o)) )
g.add( (getUrl(s), getUrl(p), getUrl(o)) )
usedTopConcepts.add( s )
usedTopConcepts.add( o )
_addUseCaseSpecificUnusedConcepts(g)
with open("hybrid-graph.rdf", "w") as f:
f.write( g.serialize() )
unusedConcepts = allTopConcepts.difference( usedTopConcepts )
print("# of unused concepts: %d" % len(unusedConcepts))
print(", ".join( list(unusedConcepts) ))
def computeOntologyStatistics(ff, cc, rc, ccCutOffCount, rcCutOffCount):
""" computes per ontology statistics (R, P, F1)
@param[in] ff list of ontology files
@param[in] cc concept counts dictionary
@param[in] rc relation counts dictionary
@param[in] ccCutOffCount min cc required for a term to be considered
@param[in] rcCutOffCount min cc required for a term to be considered
"""
goldStandardConcepts = set(
[c for c, cnt in list(cc.items()) if cnt >= ccCutOffCount])
goldStandardRelations = set(
[r for r, cnt in list(rc.items()) if cnt >= rcCutOffCount])
c = open("ontology-stats.csv", "w")
w = writer(c)
w.writerow(
("ontology", "concept precision", "concept recall", "concept F1",
"relation precision", "relation recall", "relation F1"))
for f in ff:
concepts = set(
map(str, extractConceptSet(XCL2RDF.toRDF(open(f).read()))))
relations = set(
map(str, extractRelationSet(XCL2RDF.toRDF(open(f).read()))))
cPrecision = len(goldStandardConcepts.intersection(concepts)) / float(
len(concepts))
cRecall = len(goldStandardConcepts.intersection(concepts)) / float(
len(goldStandardConcepts))
if (cPrecision + cRecall) == 0.:
cF1 = "NaN"
else:
cF1 = old_div(2 * cPrecision * cRecall, (cPrecision + cRecall))
rPrecision = len(
goldStandardRelations.intersection(relations)) / float(
len(relations))
rRecall = len(goldStandardRelations.intersection(relations)) / float(
len(goldStandardRelations))
if (rPrecision + rRecall) == 0.:
rF1 = "NaN"
else:
rF1 = old_div(2 * rPrecision * rRecall, (rPrecision + rRecall))
w.writerow((path.basename(f), cPrecision, cRecall, cF1, rPrecision,
rRecall, rF1))
c.close()
def computeStatistics( ff ):
""" computes the statistics (number of times a concept is used; number of
times a relation name is used) based on the given list of ontologies
@param[in] ff a list of files containing the ontologies to be analyzed
"""
conceptCounts = defaultdict(int)
relationCounts = defaultdict(int)
for f in ff:
concepts = set(map(str, extractConceptSet( XCL2RDF.toRDF(open(f).read() ))))
relations = set(map(str, extractRelationSet(XCL2RDF.toRDF(open(f).read() ))))
for c in concepts:
conceptCounts[c] += 1
for r in relations:
relationCounts[r] += 1
csvOutput( conceptCounts, relationCounts )
return conceptCounts, relationCounts
def computeStatistics(ff):
""" computes the statistics (number of times a concept is used; number of
times a relation name is used) based on the given list of ontologies
@param[in] ff a list of files containing the ontologies to be analyzed
"""
conceptCounts = defaultdict(int)
relationCounts = defaultdict(int)
for f in ff:
concepts = set(
map(str, extractConceptSet(XCL2RDF.toRDF(open(f).read()))))
relations = set(
map(str, extractRelationSet(XCL2RDF.toRDF(open(f).read()))))
for c in concepts:
conceptCounts[c] += 1
for r in relations:
relationCounts[r] += 1
csvOutput(conceptCounts, relationCounts)
return conceptCounts, relationCounts
def visualizeOntologies( ff ):
""" visualizes the given ontologies
@param[in] ff a list of files containing the ontologies to be visualized
"""
_createOutputDir( IMG_OUTPUT_DIR )
for f in ff:
fName, fExt = path.splitext( path.basename(f))
rdfOntology = XCL2RDF.toRDF(open(f).read() )
g = GraphvizVisualize( rdfOntology, sparqlQuery=OutputQueries._labeledGraphSparqlQuery )
g.graphTitle = fName
g.createImage( path.join(IMG_OUTPUT_DIR, fName), "pdf" )
def visualizeOntologies(ff):
""" visualizes the given ontologies
@param[in] ff a list of files containing the ontologies to be visualized
"""
_createOutputDir(IMG_OUTPUT_DIR)
for f in ff:
fName, fExt = path.splitext(path.basename(f))
rdfOntology = XCL2RDF.toRDF(open(f).read())
g = GraphvizVisualize(
rdfOntology, sparqlQuery=OutputQueries._labeledGraphSparqlQuery)
g.graphTitle = fName
g.createImage(path.join(IMG_OUTPUT_DIR, fName), "pdf")
def _readOntology( fname ):
""" reads the given ontology using the correct
format
@param[in] the ontology's file name
@returns the ontology graph
"""
if fname.endswith(".cxl"):
return XCL2RDF.toRDF( open( fname ).read() )
elif fname.endswith(".rdf") or fname.endswith(".xml"):
g = Graph()
g.parse( fname, "xml" )
return g
else:
raise "Unknown Ontology format error"
def computeOntologyStatistics( ff, cc, rc, ccCutOffCount, rcCutOffCount):
""" computes per ontology statistics (R, P, F1)
@param[in] ff list of ontology files
@param[in] cc concept counts dictionary
@param[in] rc relation counts dictionary
@param[in] ccCutOffCount min cc required for a term to be considered
@param[in] rcCutOffCount min cc required for a term to be considered
"""
goldStandardConcepts = set([ c for c, cnt in cc.items() if cnt >= ccCutOffCount ])
goldStandardRelations = set([ r for r, cnt in rc.items() if cnt >= rcCutOffCount ])
c = open("ontology-stats.csv", "w")
w = writer(c)
w.writerow( ("ontology", "concept precision", "concept recall", "concept F1",
"relation precision", "relation recall", "relation F1") )
for f in ff:
concepts = set(map(str, extractConceptSet( XCL2RDF.toRDF(open(f).read() ))))
relations = set(map(str, extractRelationSet(XCL2RDF.toRDF(open(f).read() ))))
cPrecision = len(goldStandardConcepts.intersection( concepts ))/float( len(concepts) )
cRecall = len(goldStandardConcepts.intersection( concepts ))/float( len(goldStandardConcepts) )
if (cPrecision + cRecall) == 0.:
cF1 = "NaN"
else:
cF1 = 2 * cPrecision * cRecall / (cPrecision + cRecall)
rPrecision = len(goldStandardRelations.intersection( relations ))/float( len(relations) )
rRecall = len(goldStandardRelations.intersection( relations ))/float( len(goldStandardRelations) )
if (rPrecision + rRecall) == 0.:
rF1 = "NaN"
else:
rF1 = 2 * rPrecision * rRecall / (rPrecision + rRecall)
w.writerow( (path.basename(f), cPrecision, cRecall, cF1, rPrecision, rRecall, rF1) )
c.close()
def getConcepts( fname ):
""" evaluates the given ontology and writes the results into a file
@param[in] fname file name of the ontology to evaluate
"""
goldStd = XCL2RDF.toRDF( open(fname).read() )
goldStdConcepts = extractConceptSet(goldStd)
result = []
for concept in goldStdConcepts:
cleaned_phrase = phraseCleanup.clean( concept )
if ", ".join(cleaned_phrase) != concept:
log.info("Replacing '%s' with '%s'" % (concept, ", ".join(cleaned_phrase)) )
result.extend( cleaned_phrase )
return result