def extract(txtdoc,reads,clean):
#reads = true if the document has to be read in or in other words ReadDataSet has to be run over the document
if clean:
NeoCreate.cleanDB()
if reads:
ReadDataSet.readD(txtdoc)
print("document read.")
splits = SplitText.split(txtdoc + ".ready", True)
print("document has been split.")
EntityDet.detectEnt(splits)
print("entity detection completed.")
else:
splits = SplitText.split(txtdoc, False)
print("document has been split.")
EntityDet.detectEnt(splits)
print("entity detection completed.")
NeoCreate.showAllDB()
def detectEnt(sentences):
IN = re.compile(r'.*\bin\b(?!\b.+ing)')
OF = re.compile(r'.*\bof\b(?!\b.+ing)')
#IS = re.compile(r'.*\bis\b(?!\b.+ing)(?!\b.+/VBN).*') insert IS to patterns when uncommenting
FROM = re.compile(r'.*\bfrom\b(?!\b.+ing)')
AND = re.compile(r'.*\band\b')
VB = re.compile(r'.*\b/VB\b(?!\b.+ing)(?!\b.+/VBN)') # Base form
VBD = re.compile(r'.*\b/VBD\b(?!\b.+ing)(?!\b.+/VBN)') # past tense
VBG = re.compile(r'.*\b/VBG\b') # present participle or gerund
VBN = re.compile(r'.*\b(has|have|had)(((?!been).*/VBN)|(been.*(?!\b.+/VBN)))') # has + have past participle
VBZ = re.compile(r'.*\b/VBZ\b(?!\b.+ing)(?!\b.+/VBN)') # present 3rd Sg
VBP = re.compile(r'.*\b/VBP\b(?!\b.+ing)(?!\b.+/VBN)') # present not 3rd Sg
VBPA = re.compile(r'.*\b/VBN.*\bby\b') # passive
patterns = [IN,OF,FROM,AND]
pnames = ["IN","OF","FROM","AND"]
#write document
vpatterns = [VB, VBD, VBG, VBN, VBP, VBZ,VBPA]
vnames = ["/VB", "/VBD", "/VBG","/VBN","/VBP","/VBZ","/VBN"]
f = open("testentities1", "w")
#number of relations
r = 0
print("Number of relations found: ")
for sentence in sentences:
#entity detection and parsing
sentne = nltk.ne_chunk(sentence, binary = True)
sentp = cp.parse(sentne)
#print(sentne)
ps = 0
vps = 0
for pattern in patterns:
#print("me here")
for rel in Rel.extract_rels('NE','NE', sentne, pattern, 10):
lnode = rel['subjsym']+""
rnode = rel['objsym']+""
relationship = pnames[ps]
NeoCreate.addtodb(lnode, rnode, relationship) #relation is pnames[ps]
f.write(pnames[ps] + " Relation: " + nltk.sem.relextract.show_raw_rtuple(rel) + '\n')
r += 1
print( (str(r)), end='\r')
ps+= 1
for vpattern in vpatterns:
rno = 0
for rel2 in Rel.extract_rels('NE','NE',sentne, vpattern, 10 ):
fillers = rel2['filler'].split()
verb = [v for v in fillers if v.endswith(vnames[vps])] # gets the verbs that matches the patterns
verbl = WLemmatizer.lmtz(verb[0])
if vpattern == VBPA:
ssym = rel2['subjsym']
stext = rel2 ['subjtext']
rel2['subjsym'] = rel2['objsym']
rel2['objsym'] = ssym
rel2['subjtext'] = rel2['objtext']
rel2['objtext'] = stext
lnode = rel2['subjsym']+""
rnode = rel2['objsym']+""
relationship = verbl[:(verbl.find('/'))]
NeoCreate.addtodb(lnode, rnode,relationship)
#relation is thing down there ; lnode is ssym
f.write(verbl[:(verbl.find('/'))] + " Relation: " + nltk.sem.relextract.show_raw_rtuple(rel2) + '\n')
r+=1
vps+= 1
print(str(r))
f.close
