nothing@0: import rdflib, os, fnmatch, urllib2
nothing@0: from rdflib import Graph, RDF, RDFS, plugin, URIRef, Literal, OWL, XSD, Namespace
nothing@0: from xml.dom.minidom import parseString
nothing@0: 
nothing@0: names = [line.strip() for line in open('pdfextract/names.txt')]
nothing@0: cat = [line.strip() for line in open('pdfextract/categories.txt')]
nothing@0: sig = [line.strip() for line in open('pdfextract/sig.txt')]
nothing@0: 
nothing@0: basedir = '/Users/alo/MusicOntology/features/'
nothing@0: 
nothing@0: local = 'http://sovarr.c4dm.eecs.qmul.ac.uk/features/'
nothing@0: 
nothing@0: DC = Namespace(u"http://purl.org/dc/elements/1.1/")
nothing@0: VS = Namespace(u"http://www.w3.org/2003/06/sw-vocab-status/ns#")
nothing@0: 
nothing@0: graph = Graph()
nothing@0: graph.bind('af', URIRef(local))
nothing@0: graph.bind('dc', URIRef('http://purl.org/dc/elements/1.1/'))
nothing@0: graph.bind('owl', URIRef('http://www.w3.org/2002/07/owl#'))
nothing@0: graph.bind('xsd', URIRef('http://www.w3.org/2001/XMLSchema#'))
nothing@0: graph.bind('vs', URIRef('http://www.w3.org/2003/06/sw-vocab-status/ns#'))
nothing@0: 
nothing@0: graph.add((
nothing@0:     URIRef(''),
nothing@0:     RDF.type,
nothing@0:     OWL.Ontology
nothing@0: ))
nothing@0: 
nothing@0: graph.add((
nothing@0:     URIRef(''),
nothing@0:     DC['title'],
nothing@0:     Literal("Audio Features Base Ontology")
nothing@0: ))
nothing@0: 
nothing@0: graph.add((
nothing@0:     URIRef(''),
nothing@0:     OWL.versionInfo,
nothing@0:     Literal("Version 0.1")
nothing@0: ))
nothing@0: 
nothing@0: graph.add((
nothing@0:     URIRef(''),
nothing@0:     DC['description'],
nothing@0:     Literal("This is a base ontology for the Audio Features engineering process collected from literature")
nothing@0: ))
nothing@0: 
nothing@0: graph.add((
nothing@0:     VS['term_status'],
nothing@0:     RDF.type,
nothing@0:     OWL.AnnotationProperty
nothing@0: ))
nothing@0: 
nothing@0: i = 0
nothing@0: 
nothing@0: order = [
nothing@0:     "Zero Crossing Rate", 
nothing@0:     "Linear Predictive Coding", 
nothing@0:     "Mel-scale Frequency Cepstral Coefficients", 
nothing@0:     "Auditory Filter Bank Temporal Envelopes",
nothing@0:     "Rate-scale-frequency Features",
nothing@0:     "Phase Space Features"
nothing@0: ]
nothing@0: 
nothing@0: domains = {
nothing@0:     "Zero Crossing Rate": 'temporal',
nothing@0:     "Linear Predictive Coding": 'frequency',
nothing@0:     "Mel-scale Frequency Cepstral Coefficients": 'cepstral',
nothing@0:     "Auditory Filter Bank Temporal Envelopes": 'modulation frequency',
nothing@0:     "Rate-scale-frequency Features": 'eigendomain',
nothing@0:     "Phase Space Features": 'phase space'
nothing@0: }
nothing@0: 
nothing@0: abbr = {
nothing@0: 	"Zero Crossing Rate": "ZCR",
nothing@0: 	"Mel-scale Frequency Cepstral Coefficients": "MFCC",
nothing@0:     "Linear Predictive Coding": "LPC",
nothing@0:     "Linear Prediction Cepstral Coefficients": "LPCC",
nothing@0:     "Zero crossing peak amplitudes": "ZCPA",
nothing@0:     "Line spectral frequencies": "LSF",
nothing@0:     "Short-time energy": "STE",
nothing@0:     "Amplitude descriptor":  "AD",
nothing@0:     "Adaptive time frequency transform": "ATFT",
nothing@0:     "Daubechies Wavelet coefficient histogram": "DWCH",
nothing@0:     "Spectral Flux": "SF",
nothing@0:     "Group delay function": "GDF",
nothing@0:     "Modified group delay function": "MGDF",
nothing@0:     "Spectral centroid": "SC",
nothing@0:     "Subband spectral flux": "SSF",
nothing@0:     "Perceptual linear prediction": "PLP"
nothing@0: }
nothing@0: 
nothing@0: appdom = {
nothing@0:     'ASR': "Speech Recognition",
nothing@0:     'ESR': "Environmental Sound Recognition",
nothing@0:     'MIR': "Music Information Retrieval",
nothing@0:     'AS': "Audio Segmentation",
nothing@0:     'FP': "Fingerprinting",
nothing@0:     'VAR': "Several",
nothing@0:     'EXC': ""
nothing@0: }
nothing@0: 
nothing@0: domain = ""
nothing@0: domainIndex = 0
nothing@0: compdict = {}
nothing@0: 
nothing@0: graph.add((
nothing@0:     URIRef(local + 'MathematicalOperation'),
nothing@0:     RDF.type,
nothing@0:     OWL.Class
nothing@0: ))
nothing@0: 
nothing@0: graph.add((
nothing@0:     URIRef(local + 'Filter'),
nothing@0:     RDF.type,
nothing@0:     OWL.Class
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + 'Filter'),
nothing@0:     RDFS.subClassOf,
nothing@0:     URIRef(local + 'MathematicalOperation')
nothing@0: ))
nothing@0: 
nothing@0: graph.add((
nothing@0:     URIRef(local + 'Transformation'),
nothing@0:     RDF.type,
nothing@0:     OWL.Class
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + 'Transformation'),
nothing@0:     RDFS.subClassOf,
nothing@0:     URIRef(local + 'MathematicalOperation')
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + 'Aggregation'),
nothing@0:     RDF.type,
nothing@0:     OWL.Class
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + 'Aggregation'),
nothing@0:     RDFS.subClassOf,
nothing@0:     URIRef(local + 'MathematicalOperation')
nothing@0: ))
nothing@0: 
nothing@0: for filename in ['filters', 'trans', 'aggr']:
nothing@0:     compsuper = filename.replace('filters', 'Filter').replace('trans', 'Transformation').replace('aggr', 'Aggregation')
nothing@0:     for line in [line.strip() for line in open(basedir + 'pdfextract/' + filename + '.txt')]:
nothing@0:         compname = line[2:]
nothing@0:         compidref = URIRef(local + compname.replace(' ', '').replace('(', '').replace(')', '').replace('-', '').replace(',', ''))
nothing@0:         graph.add((
nothing@0:             compidref,
nothing@0:             RDF.type,
nothing@0:             OWL.Class
nothing@0:         ))
nothing@0:         graph.add((
nothing@0:             compidref,
nothing@0:             RDFS.subClassOf,
nothing@0:             URIRef(local + compsuper)
nothing@0:         ))
nothing@0:         graph.add((
nothing@0:             compidref,
nothing@0:             RDFS.label,
nothing@0:             Literal(compname)
nothing@0:         ))
nothing@0:         compdict[line[0]] = compidref
nothing@0: 
nothing@0: graph.add((
nothing@0:     URIRef(local + 'Signal'),
nothing@0:     RDF.type, 
nothing@0:     OWL.Class 
nothing@0: ))
nothing@0: 
nothing@0: graph.add((
nothing@0:     URIRef(local + 'Feature'),
nothing@0:     RDF.type, 
nothing@0:     OWL.Class 
nothing@0: ))
nothing@0: 
nothing@0: graph.add((
nothing@0:     URIRef(local + 'Feature'),
nothing@0:     OWL.subClassOf, 
nothing@0:     URIRef(local + 'Signal'),
nothing@0: ))
nothing@0: 
nothing@0: for dom in domains.values():
nothing@0:     idref = URIRef(local + dom.capitalize().replace(' ', '') + 'Feature')
nothing@0:     graph.add((
nothing@0:         idref,
nothing@0:         RDF.type, 
nothing@0:         OWL.Class  
nothing@0:     ))
nothing@0:     graph.add((
nothing@0:         idref,
nothing@0:         RDFS.subClassOf, 
nothing@0:         URIRef(local + 'Feature')
nothing@0:     ))
nothing@0: 
nothing@0: graph.add((
nothing@0:     URIRef(local + 'PerceptualFeature'),
nothing@0:     RDF.type, 
nothing@0:     OWL.Class
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + 'PerceptualFeature'),
nothing@0:     RDFS.subClassOf, 
nothing@0:     URIRef(local + 'Feature')
nothing@0: ))
nothing@0: 
nothing@0: graph.add((
nothing@0:     URIRef(local + 'FrequencyDomainPerceptualFeature'),
nothing@0:     RDF.type, 
nothing@0:     OWL.Class
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + 'FrequencyDomainPerceptualFeature'),
nothing@0:     RDFS.subClassOf, 
nothing@0:     URIRef(local + 'FrequencyFeature')
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + 'FrequencyDomainPerceptualFeature'),
nothing@0:     OWL.equivalentClass, 
nothing@0:     URIRef(local + 'PerceptualFeature')
nothing@0: ))
nothing@0: 
nothing@0: graph.add((
nothing@0:     URIRef(local + 'FrequencyDomainPhysicalFeature'),
nothing@0:     RDF.type, 
nothing@0:     OWL.Class
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + 'FrequencyDomainPhysicalFeature'),
nothing@0:     RDFS.subClassOf, 
nothing@0:     URIRef(local + 'FrequencyFeature')
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + 'FrequencyDomainPhysicalFeature'),
nothing@0:     OWL.equivalentClass, 
nothing@0:     URIRef(local + 'PhysicalFeature')
nothing@0: ))
nothing@0: 
nothing@0: 
nothing@0: 
nothing@0: graph.add((
nothing@0:     URIRef(local + 'PhysicalFeature'),
nothing@0:     RDF.type, 
nothing@0:     OWL.Class  
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + 'PhysicalFeature'),
nothing@0:     RDFS.subClassOf, 
nothing@0:     URIRef(local + 'Feature')
nothing@0: ))
nothing@0: 
nothing@0: graph.add((
nothing@0:     URIRef(local + 'ParametrizedDimensions'),
nothing@0:     RDF.type, 
nothing@0:     OWL.Class  
nothing@0: ))
nothing@0: 
nothing@0: graph.add((
nothing@0:     URIRef(local + 'ComputationalComplexity'),
nothing@0:     RDF.type, 
nothing@0:     OWL.Class  
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + 'LowComplexity'),
nothing@0:     RDF.type, 
nothing@0:     OWL.Class  
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + 'LowComplexity'),
nothing@0:     RDFS.subClassOf, 
nothing@0:     URIRef(local + 'ComputationalComplexity')
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + 'MediumComplexity'),
nothing@0:     RDF.type, 
nothing@0:     OWL.Class  
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + 'MediumComplexity'),
nothing@0:     RDFS.subClassOf, 
nothing@0:     URIRef(local + 'ComputationalComplexity')
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + 'HighComplexity'),
nothing@0:     RDF.type, 
nothing@0:     OWL.Class  
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + 'HighComplexity'),
nothing@0:     RDFS.subClassOf, 
nothing@0:     URIRef(local + 'ComputationalComplexity')
nothing@0: ))
nothing@0: 
nothing@0: graph.add((
nothing@0:     URIRef(local + 'TemporalScale'),
nothing@0:     RDF.type, 
nothing@0:     OWL.Class  
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + 'IntraFrame'),
nothing@0:     RDF.type, 
nothing@0:     OWL.Class  
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + 'IntraFrame'),
nothing@0:     RDFS.subClassOf, 
nothing@0:     URIRef(local + 'TemporalScale')
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + 'InterFrame'),
nothing@0:     RDF.type, 
nothing@0:     OWL.Class  
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + 'InterFrame'),
nothing@0:     RDFS.subClassOf, 
nothing@0:     URIRef(local + 'TemporalScale')
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + 'Global'),
nothing@0:     RDF.type, 
nothing@0:     OWL.Class  
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + 'Global'),
nothing@0:     RDFS.subClassOf, 
nothing@0:     URIRef(local + 'TemporalScale')
nothing@0: ))
nothing@0: 
nothing@0: 
nothing@0: graph.add((
nothing@0:     URIRef(local + 'ApplicationDomain'),
nothing@0:     RDF.type, 
nothing@0:     OWL.Class  
nothing@0: ))
nothing@0: 
nothing@0: for key in appdom.keys():
nothing@0:     if appdom[key] != "":
nothing@0:         idref = URIRef(local + appdom[key].replace(" ", ""))
nothing@0:         graph.add((
nothing@0:             idref,
nothing@0:             URIRef(RDF.type), 
nothing@0:             OWL.Class  
nothing@0:         ))
nothing@0:         graph.add((
nothing@0:             idref,
nothing@0:             RDFS.subClassOf, 
nothing@0:             URIRef(local + 'ApplicationDomain')
nothing@0:         ))
nothing@0: 
nothing@0: #properties
nothing@0: graph.add((
nothing@0:     URIRef(local + "application_domain"),
nothing@0:     RDF.type,
nothing@0:     RDF.Property
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + "application_domain"),
nothing@0:     RDFS.range,
nothing@0:     URIRef(local + 'ApplicationDomain')
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + "application_domain"),
nothing@0:     VS['term_status'],
nothing@0:     Literal("testing")
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + "application_domain"),
nothing@0:     RDFS.comment,
nothing@0:     Literal("application domain property")
nothing@0: ))
nothing@0: 
nothing@0: 
nothing@0: 
nothing@0: graph.add((
nothing@0:     URIRef(local + "semantic_interpretation"),
nothing@0:     RDF.type,
nothing@0:     RDF.Property
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + "semantic_interpretation"),
nothing@0:     VS['term_status'],
nothing@0:     Literal("testing")
nothing@0: ))
nothing@0:     
nothing@0: graph.add((
nothing@0:     URIRef(local + "computational_complexity"),
nothing@0:     RDF.type,
nothing@0:     RDF.Property
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + "computational_complexity"),
nothing@0:     VS['term_status'],
nothing@0:     Literal("testing")
nothing@0: ))
nothing@0: 
nothing@0: graph.add((
nothing@0:     URIRef(local + "computational_complexity"),
nothing@0:     RDFS.range,
nothing@0:     URIRef(local + 'ComputationalComplexity')
nothing@0: ))
nothing@0: 
nothing@0: graph.add((
nothing@0:     URIRef(local + "psychoacoustic_model"),
nothing@0:     RDF.type,
nothing@0:     RDF.Property
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + "psychoacoustic_model"),
nothing@0:     RDFS.range,
nothing@0:     XSD.Boolean
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + "psychoacoustic_model"),
nothing@0:     VS['term_status'],
nothing@0:     Literal("testing")
nothing@0: ))
nothing@0: 
nothing@0: 
nothing@0: graph.add((
nothing@0:     URIRef(local + "dimensions"),
nothing@0:     RDF.type,
nothing@0:     RDF.Property
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + "dimensions"),
nothing@0:     RDFS.range,
nothing@0:     XSD.Integer
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + "dimensions"),
nothing@0:     RDFS.range,
nothing@0:     URIRef(local + 'ParametrizedDimensions')
nothing@0: ))
nothing@0: 
nothing@0: graph.add((
nothing@0:     URIRef(local + "temporal_scale"),
nothing@0:     RDF.type,
nothing@0:     RDF.Property
nothing@0: ))
nothing@0: graph.add((
nothing@0:     URIRef(local + "temporal_scale"),
nothing@0:     RDFS.range,
nothing@0:     URIRef(local + 'TemporalScale')
nothing@0: ))
nothing@0:             
nothing@0: for name in names:
nothing@0:     id = local + (name.replace(' ','').replace('-',''))
nothing@0:     
nothing@0:     if name == order[domainIndex]:
nothing@0:         domain = domains[order[domainIndex]]
nothing@0:         domainIndex += 1
nothing@0:     
nothing@0:     graph.add(( URIRef(id),        
nothing@0:         URIRef('http://www.w3.org/1999/02/22-rdf-syntax-ns#type'), 
nothing@0:         OWL.Class  
nothing@0:     ))
nothing@0:     
nothing@0:     graph.add((
nothing@0:         URIRef(id),
nothing@0:         VS['term_status'],
nothing@0:         Literal("testing")
nothing@0:     ))
nothing@0: 
nothing@0:     if domain == "frequency":
nothing@0:         if word[1] == 'Y': 
nothing@0:             temp = URIRef(local + 'FrequencyDomainPerceptualFeature')
nothing@0:         else:
nothing@0:             temp = URIRef(local + 'FrequencyDomainPhysicalFeature')
nothing@0: 
nothing@0:         graph.add(( 
nothing@0:             URIRef(id), 
nothing@0:             RDFS.subClassOf, 
nothing@0:             URIRef(temp)
nothing@0:         ))
nothing@0:             
nothing@0:     else:
nothing@0:         graph.add(( 
nothing@0:             URIRef(id), 
nothing@0:             RDFS.subClassOf, 
nothing@0:             URIRef(local + domain.capitalize().replace(' ', '') + 'Feature')
nothing@0:         ))
nothing@0:     
nothing@0:     graph.add(( 
nothing@0:         URIRef(id), 
nothing@0:         #URIRef(local + 'feature'), 
nothing@0:         RDFS.label,
nothing@0:         Literal(name.replace(' ','').replace('-','')) 
nothing@0:     ))
nothing@0: 
nothing@0:     graph.add(( 
nothing@0:         URIRef(id), 
nothing@0:         RDFS.comment,
nothing@0:         Literal(name + " feature") 
nothing@0:     ))
nothing@0: 
nothing@0:     graph.add(( 
nothing@0:         URIRef(id), 
nothing@0:         RDFS.label, 
nothing@0:         Literal(name) 
nothing@0:     ))
nothing@0: 
nothing@0:     word = cat[i].split(' ')
nothing@0: 
nothing@0:     temp = {
nothing@0:         'I': URIRef(local+'IntraFrame'),
nothing@0:         'X': URIRef(local+'InterFrame'),
nothing@0:         'G': URIRef(local+'Global')
nothing@0:     }[word[0]]
nothing@0: 
nothing@0:     graph.add(( 
nothing@0:         URIRef(id), 
nothing@0:         URIRef(local + 'temporal_scale'), 
nothing@0:         temp 
nothing@0:     ))
nothing@0:     
nothing@0:     
nothing@0:     if word[1] == 'Y': 
nothing@0:         temp = URIRef(local + 'PerceptualFeature')
nothing@0:     else:
nothing@0:         temp = URIRef(local + 'PhysicalFeature')
nothing@0:         
nothing@0:     graph.add((
nothing@0:         URIRef(id), 
nothing@0:         URIRef(local + "semantic_interpretation"), 
nothing@0:         temp
nothing@0:     ))
nothing@0:         
nothing@0:     if word[2] == 'Y':
nothing@0:         graph.add((
nothing@0:             URIRef(id), 
nothing@0:             URIRef(local + "psychoacoustic_model"), 
nothing@0:             Literal(True) 
nothing@0:         ))
nothing@0:     else:
nothing@0:         graph.add((
nothing@0:             URIRef(id), 
nothing@0:             URIRef(local + "psychoacoustic_model"), 
nothing@0:             Literal(False) 
nothing@0:         ))
nothing@0:         
nothing@0:     temp = {
nothing@0:         'L': URIRef(local + 'LowComplexity'),
nothing@0:         'M': URIRef(local + 'MediumComplexity'),
nothing@0:         'H': URIRef(local + 'HighComplexity')
nothing@0:     }[word[3]]
nothing@0:     
nothing@0:     graph.add(( 
nothing@0:         URIRef(id),
nothing@0:         URIRef(local + "computational_complexity"), 
nothing@0:         temp 
nothing@0:     ))
nothing@0:     
nothing@0:     if word[4] == 'V':
nothing@0:         temp = URIRef(local + 'ParametrizedDimensions')
nothing@0:     else:
nothing@0:         temp = Literal(int(word[4]))
nothing@0:         
nothing@0:     graph.add(( 
nothing@0:         URIRef(id),
nothing@0:         URIRef(local + 'dimensions'), 
nothing@0:         temp
nothing@0:     ))
nothing@0:     
nothing@0:     temp = appdom[word[5]]
nothing@0: 
nothing@0:     if temp != '':   
nothing@0:         graph.add(( 
nothing@0:             URIRef(id), 
nothing@0:             URIRef(local + "application_domain"),            
nothing@0:             URIRef(local + temp.replace(" ", "")) 
nothing@0:         ))            
nothing@0:         
nothing@0:     steps = sig[i].split(' ')
nothing@0:     
nothing@0:     for key in steps:
nothing@0:         graph.add(( 
nothing@0:             URIRef(id), 
nothing@0:             URIRef(local + 'computation'), 
nothing@0:             compdict[key] 
nothing@0:         ))
nothing@0:         
nothing@0:     if name.find('MPEG-7') >= 0:
nothing@0:         graph.add(( 
nothing@0:             URIRef(id),
nothing@0:             URIRef(local + 'computedIn'),
nothing@0:             Literal('MPEG-7')
nothing@0:         ))
nothing@0:         #graph.add(( 
nothing@0:         #    URIRef(local+name.replace('MPEG-7', '').lower().lstrip().replace(' ', '_')+'_feature'),
nothing@0:         #    RDF.type,
nothing@0:         #    URIRef(id)
nothing@0:         #))
nothing@0:     
nothing@0:     if name in abbr.keys():
nothing@0:         graph.add(( 
nothing@0:             URIRef(id),
nothing@0:             URIRef(local + 'abbreviation'),
nothing@0:             Literal(abbr[name])
nothing@0:         ))
nothing@0:         
nothing@0:     
nothing@0:     i += 1
nothing@0: 
nothing@0: 
nothing@0: 
nothing@0: graph.serialize('/Users/alo/MusicOntology/features/baseOnto.n3', format='n3')
nothing@0: graph.serialize('/Users/alo/MusicOntology/features/baseOnto.rdf')