def get_collocations_alchemy(self):
"""Extract collocations using the self.cleaned_data dictionary"""
full_name = handle_uploaded_file(self.cleaned_data['article_pdf'])
if not os.path.exists(full_name + "x.xml"):
subprocess.call([settings.PDFX_PATH, full_name])
extracted_data = parse_pdfx_xml(full_name + "x.xml")
full_text = nlp.get_full_text(extracted_data)['text']
payload = {'apikey': 'd0604109bbeb676474b243bc623a0fc1a172437f', 'outputMode': 'json',
'maxRetrieve': '100', 'text': full_text}
response = requests.post('http://access.alchemyapi.com/calls/text/TextGetRankedNamedEntities',
data=payload)
result = response.json()
labels = []
for entity in result['entities']:
labels.append((entity['text'] + ' (' + entity['type'] + ')', entity['relevance']))
return labels
def get_collocations(self):
"""Extract collocations using the self.cleaned_data dictionary"""
from axel.stats.models import Collocations
full_name = handle_uploaded_file(self.cleaned_data['article_pdf'])
#stem_func = getattr(Stemmer, self.cleaned_data['stem_func'])
if not os.path.exists(full_name + "x.xml"):
subprocess.call([settings.PDFX_PATH, full_name])
extracted_data = parse_pdfx_xml(full_name + "x.xml")
full_text = nlp.get_full_text(extracted_data)['text']
article = PDFUploadForm.generate_temp_article(full_text)
labels = []
try:
features = PDFUploadForm.build_features(article)
for ngram, feature in features:
klass0 = self.CLF.predict_proba(feature)[0][1]
labels.append((ngram, klass0))
finally:
article.delete()
Collocations.objects.filter(count=0).delete()
labels.sort(key=lambda x: x[1], reverse=True)
return labels