def hasIntensifierFeature(procTxt, hr, featureVals = {}):
featureVals = hasIntensifier(procTxt, hr, featureVals)
intense_value = featureVals['hasIntensifier']
featureSets = {}
intensifierdict ={}
intensifierdict.update(discVar2Feature(intense_value, 'hasIntensifier', lims = [1,1],collapse = [False, True]))
featureSets.update(intensifierdict)
return (featureVals, featureSets)
def hasIntensifierFeature(procTxt, hr, featureVals = {}):
featureVals = hasIntensifier(procTxt, hr, featureVals)
intense_value = featureVals['hasIntensifier']
featureSets = {}
intensifierdict ={}
intensifierdict.update(discVar2Feature(intense_value, 'hasIntensifier', lims = [1,1],collapse = [False, True]))
featureSets.update(intensifierdict)
return (featureVals, featureSets)
def countSmileyFeature(tokTag, hr, featureVals = {}):
"""
count Smiley Feature Encoding
3 level discretization for
"""
csname = 'countSmiley'
nsname = 'negative smiley'
psname = 'positive smiley'
featureVals = countSmiley(tokTag, hr, featureVals)
np, nn = featureVals[csname]
nfdict = discVar2Feature(featureVals[csname][0], nsname, lims = [1,1]) #negative smiley
pfdict = discVar2Feature(featureVals[csname][1], psname, lims = [1,1]) #positive smiley
featureSets = {}
featureSets.update(pfdict.items())
featureSets.update(nfdict.items())
return (featureVals, featureSets)
def countSmileyFeature(tokTag, hr, featureVals = {}):
"""
count Smiley Feature Encoding
3 level discretization for
"""
csname = 'countSmiley'
nsname = 'negative smiley'
psname = 'positive smiley'
featureVals = countSmiley(tokTag, hr, featureVals)
np, nn = featureVals[csname]
nfdict = discVar2Feature(featureVals[csname][0], nsname, lims = [1,1]) #negative smiley
pfdict = discVar2Feature(featureVals[csname][1], psname, lims = [1,1]) #positive smiley
featureSets = {}
featureSets.update(pfdict.items())
featureSets.update(nfdict.items())
return (featureVals, featureSets)
def multiverb_sentFeature(tokTag, hr, featureVals = {}):
featureVals = multiverb_sent(tokTag, hr, featureVals)
fv = featureVals['multiverb_sent']
retval = {}
retval.update(discVar2Feature(fv, 'multiverbsentences', lims = [1,1], collapse = [False, True]))
#retval['multiverb_sent'] = fv
featureSets = {}
featureSets.update(retval)
return (featureVals, featureSets)
def multiverb_sentFeature(tokTag, hr, featureVals = {}):
featureVals = multiverb_sent(tokTag, hr, featureVals)
fv = featureVals['multiverb_sent']
retval = {}
retval.update(discVar2Feature(fv, 'multiverbsentences', lims = [1,1], collapse = [False, True]))
#retval['multiverb_sent'] = fv
featureSets = {}
featureSets.update(retval)
return (featureVals, featureSets)
def countPosteriorPolarChunksFeature(procTxt, hr, featureVals = {}):
featureVals = countPosteriorPolarChunks(procTxt, hr, featureVals)
cnt = featureVals['countPosteriorPolarChunks']
retdict = dict()
for key, val in cnt.iteritems():
retdict.update(discVar2Feature(val, key, lims = [1,3], collapse = [False, True]))
featureSets = {}
featureSets.update(retdict)
return (featureVals, featureSets)
def netPosteriorPolarChunksFeature(procTxt, hr, featureVals = {}):
featureVals = netPosteriorPolarChunks(procTxt, hr, featureVals)
cnt = featureVals['netPosteriorPolarChunks']
#print cnt
retdict = dict()
retdict.update(discVar2Feature(cnt, 'netPosteriorPolarChunks', lims = [-1,3], collapse = [False, True]))
featureSets = {}
featureSets.update(retdict) #{'NPP':cnt})
return (featureVals, featureSets)
def countPosteriorPolarChunksFeature(procTxt, hr, featureVals = {}):
featureVals = countPosteriorPolarChunks(procTxt, hr, featureVals)
cnt = featureVals['countPosteriorPolarChunks']
retdict = dict()
for key, val in cnt.iteritems():
retdict.update(discVar2Feature(val, key, lims = [1,3], collapse = [False, True]))
featureSets = {}
featureSets.update(retdict)
return (featureVals, featureSets)
def netPosteriorPolarChunksFeature(procTxt, hr, featureVals = {}):
featureVals = netPosteriorPolarChunks(procTxt, hr, featureVals)
cnt = featureVals['netPosteriorPolarChunks']
#print cnt
retdict = dict()
retdict.update(discVar2Feature(cnt, 'netPosteriorPolarChunks', lims = [-1,3], collapse = [False, True]))
featureSets = {}
featureSets.update(retdict) #{'NPP':cnt})
return (featureVals, featureSets)
def countPolarChunksFeature(procTxt, hr, featureVals = {}):
featureVals = countPolarChunks(procTxt, hr, featureVals)
cnt = featureVals['countPolarChunks']
retdict = dict()
for chtype in TPC_CHUNKTYPES:
for pol in TPC_POLTYPES:
key = chtype+pol
retdict.update(discVar2Feature(cnt[key], key, lims = [1,3], collapse = [False, True]))
featureSets = {}
featureSets.update(retdict)
return (featureVals, featureSets)
def countPolarChunksFeature(procTxt, hr, featureVals = {}):
featureVals = countPolarChunks(procTxt, hr, featureVals)
cnt = featureVals['countPolarChunks']
retdict = dict()
for chtype in TPC_CHUNKTYPES:
for pol in TPC_POLTYPES:
key = chtype+pol
retdict.update(discVar2Feature(cnt[key], key, lims = [1,3], collapse = [False, True]))
featureSets = {}
featureSets.update(retdict)
return (featureVals, featureSets)
def netPolarChunksFeature(procTxt, hr, featureVals = {}):
featureVals = netPolarChunks(procTxt, hr, featureVals)
cnt = featureVals['netPolarChunks']
#print cnt
retdict = dict()
retdict.update(discVar2Feature(cnt, 'netPolarChunks', lims = [-1,1], collapse = [False, True]))
# for chtype in TPC_CHUNKTYPES:
# for pol in TPC_POLTYPES:
# key = chtype+pol
# retdict.update(discVar2Feature(cnt[key], key, lims = [1,3], collapse = [False, True]))
featureSets = {}
featureSets.update(retdict)
return (featureVals, featureSets)
def netPolarChunksFeature(procTxt, hr, featureVals = {}):
featureVals = netPolarChunks(procTxt, hr, featureVals)
cnt = featureVals['netPolarChunks']
#print cnt
retdict = dict()
retdict.update(discVar2Feature(cnt, 'netPolarChunks', lims = [-1,1], collapse = [False, True]))
# for chtype in TPC_CHUNKTYPES:
# for pol in TPC_POLTYPES:
# key = chtype+pol
# retdict.update(discVar2Feature(cnt[key], key, lims = [1,3], collapse = [False, True]))
featureSets = {}
featureSets.update(retdict)
return (featureVals, featureSets)
def countNGInTelecomDictFeature(tokTag, hr, featureVals = {}):
"""
Telecom Dict feature
"""
#featureVals = countNGInTelecomDict(tokTag, hr, featureVals)
featureVals = countNGInDomainDict(tokTag, hr, featureVals)
ncount = featureVals['countNGInDomainDict']
ngname = 'NG In TELCOM Dict'
lims = [1,3]
collapse = [False, True]
featureSets = {}
for k in ncount:
tdict = discVar2Feature(len(ncount[k]), k + ngname, lims = lims, collapse = collapse) #negative smiley
featureSets.update(tdict.items())
return (featureVals, featureSets)
def countNGInTelecomDictFeature(tokTag, hr, featureVals = {}):
"""
Telecom Dict feature
"""
#featureVals = countNGInTelecomDict(tokTag, hr, featureVals)
featureVals = countNGInDomainDict(tokTag, hr, featureVals)
ncount = featureVals['countNGInDomainDict']
ngname = 'NG In TELCOM Dict'
lims = [1,3]
collapse = [False, True]
featureSets = {}
for k in ncount:
tdict = discVar2Feature(len(ncount[k]), k + ngname, lims = lims, collapse = collapse) #negative smiley
featureSets.update(tdict.items())
return (featureVals, featureSets)
def countPolarPOSFeature(tokTag, hr, featureVals = {}):
"""
count Polar POS Feature Encoding
5 level discretization
"""
featureVals = countPolarPOS(tokTag, hr, featureVals)
cnt = featureVals['countPolarPOS']
#work on this should not have to deal with labels here....
senlbls = ['negative', 'positive']
POSLbls = ['adjective', 'adverb', 'verb'] #,'noun','interjection']
retdict = dict()
for pos in POSLbls:
for sen in senlbls:
retdict.update(discVar2Feature(cnt[sen][pos], sen +' '+ pos, lims = [1,3], collapse = [False, True]))
#retdict.update(discVar2Feature(cnt[sen][pos], sen +' '+ pos, lims = [1,5], collapse = [False, True]))
featureSets = {}
featureSets.update(retdict)
return (featureVals, featureSets)
def countPolarPOSFeature(tokTag, hr, featureVals = {}):
"""
count Polar POS Feature Encoding
5 level discretization
"""
featureVals = countPolarPOS(tokTag, hr, featureVals)
cnt = featureVals['countPolarPOS']
#work on this should not have to deal with labels here....
senlbls = ['negative', 'positive']
POSLbls = ['adjective', 'adverb', 'verb'] #,'noun','interjection']
retdict = dict()
for pos in POSLbls:
for sen in senlbls:
retdict.update(discVar2Feature(cnt[sen][pos], sen +' '+ pos, lims = [1,3], collapse = [False, True]))
#retdict.update(discVar2Feature(cnt[sen][pos], sen +' '+ pos, lims = [1,5], collapse = [False, True]))
featureSets = {}
featureSets.update(retdict)
return (featureVals, featureSets)