def parseAllSentencePfm(self): #list them all, becaue if loop with cursor and update cursor pointed sentence at meantime, the cursor will be screwed. sentences = list(self.db.getAllSentence()) for i, sentence in enumerate(sentences): print(i) pfmSentenceWordList = getProcessedWordList(sentence['content'], NOUN) pfmWordList = filter(lambda word : word in self.pfmWord, pfmSentenceWordList) posNegSentenceWordList = getProcessedWordList(sentence['content'], VERB) posWordList = filter(lambda word : word in self.posWord, posNegSentenceWordList) negWordList = filter(lambda word : word in self.negWord, posNegSentenceWordList) posWordList, negWordList = self.filterPosNegWordListByDistance(pfmSentenceWordList, posNegSentenceWordList, pfmWordList, posWordList, negWordList) self.db.updateSentencePfm(sentence['_id'], pfmWordList, posWordList, negWordList)
def isCiteInDistance(self, sentence): #if (CEO or Company) and citation word happens within 5 word distance, capture isCiteCEO, isCiteAnalyst, isCiteCompany = False, False, False if sentence['cite']: wordList = getProcessedWordList(sentence['content'], VERB) for citeWord in sentence['cite']: citeIndex = wordList.index(citeWord) for engagerId in sentence['engager']: try: engager = self.db.getEngagerById(engagerId) matchName = engager['lastName'].lower() engagerIndex = wordList.index(matchName) if abs(citeIndex - engagerIndex) <= CITE_DISTANCE: if engager['type'] == ENGAGER_CEO: isCiteCEO = True else: isCiteAnalyst = True except: pass for companyId in sentence['company']: try: company = self.db.getCompanyById(companyId) matchName = company['shortName'].lower() companyIndex = wordList.index(matchName) if abs(citeIndex - companyIndex) <= CITE_DISTANCE: isCiteCompany = True except: pass return isCiteCEO, isCiteAnalyst, isCiteCompany
def parseAllSentenceCitation(self): sentences = list(self.db.getAllSentence()) for i, sentence in enumerate(sentences): print(i) words = getProcessedWordList(sentence['content'], VERB) sentence['cite'] = filter(lambda word : word in self.citeWord, words) sentence['citeCEO'], sentence['citeAnalyst'], sentence['citeCompany'] = self.isCiteInDistance(sentence) self.db.saveSentence(sentence)
def parseAllSentenceAtrb(self):
sentences = list(self.db.getAllSentence())
for i, sentence in enumerate(sentences):
print(i)
words = getProcessedWordList(sentence['content'], NOUN)
exWordList = filter(lambda word : word in self.exWord, words)
inWordList = filter(lambda word : word in self.inWord, words)
if ('ceo' in inWordList or 'executive' in inWordList) and sentence['cite']:
inWordList = []
self.db.updateSentenceAtrb(sentence['_id'], exWordList, inWordList)
def getSentenceMatrixAndIdList(self, sentences):
table = self.db.getUnigramTable()
matrix, idList = [], []
i = 0
for sentence in sentences:
print i
i += 1
wordList = getProcessedWordList(sentence['content'])
vector = self.getSentenceVector(table, wordList)
matrix.append(vector)
idList.append(sentence['_id'])
return numpy.array(matrix), idList
