def docToclassvector(self, document, alltopicsandkeys):
result = alltopicsandkeys #self.traversePages("setTopicandTerms",'topicmodel')
classifier = Classifier()
classifier.init(result.topics, result.terms).MinKey(2)
classifier.build(document)
classifier.setweights()
return classifier
def getTopics(self):
result = self.traversePages("setTopicandTerms", 'topicmodel')
classifier = Classifier()
classifier.init(result.topics, result.terms).MinKey(2)
classifier = self.traversePages("retrainClassifier", 'article',
classifier).classifier
classifier.setweights()
os.remove("classvectors.json")
classifier.tojson("classvectors")
return "test2"
def getCategory(self, query, option, alltopicsandkeys):
result = alltopicsandkeys
outcome = {}
merger = Merger()
classifier = Classifier()
# result = self.traversePages("setTopicandTerms",'topicmodel')
classifier.init(result.topics, result.terms).MinKey(2)
# classifier.load('articlemodel.json')
classifier.load('classvectors.json')
# queryvector.tojson('query')
# print(queryvector.model['model'])
outcome['categoriesconfidence'] = {}
alreadymerged = 0
for k, arr in result.terms.items():
value = classifier.goodtopicscore(arr, query.lower())
outcome['categoriesconfidence'][k] = value
if option == 1:
if value > 0 and not alreadymerged:
queryvector = self.docToclassvector(query, result)
classvectormodels = []
classvectormodels.append(classifier)
classvectormodels.append(queryvector)
classifier = merger.merge(classvectormodels)
alreadymerged = 1
# print("{} {}".format(k,value))
value = 0
classifier.tojson("classvectors")
poutcome = classifier.predict('model', query).getTopics()
k = Counter(poutcome)
# Finding 3 highest values
answer = k.most_common(3)
categories = poutcome
outcome['categoriestop3'] = answer
outcome['categories'] = categories
outcome['document'] = query
outcome['categorieswordmatch'] = {}
for k, v in classifier.termVectors.items():
outcome['categorieswordmatch'][k] = str(v)
# print(k)
# print(v)
# print(type(classifier.termVectors))
# print(classifier.termVectors)
return json.dumps(outcome)