Python CCorpusManager Examples

Example #1

File: index.py Project: almightyGOSU/CZ4045-NLP-King-Chody-Almighty-Backend

def getTypes():
	strOutput = ""
	for strWord in CCorpusManager.getTokenList():
		strOutput += strWord
		strOutput += "<br/>"

	return strOutput

Example #2

File: CSourceManager.py Project: almightyGOSU/CZ4045-NLP-King-Chody-Almighty-Backend

def getSourceSimilarity(pStrWord):

	strText = ""
	lstTokens = []

	for intCount in range(1, CCorpusManager.getDocumentsCount() + 1):
		lstTokens += word_tokenize(CFileManager.readFromFile(str(intCount)))

	return lstTokens

	#objCI = nltk.text.ContextIndex([word.lower() for word in lstTokens])
	#objCI = nltk.text.ContextIndex(['tasty','fluffy','yummy','','','','',''])
	#return objCI.similar_words(pStrWord)

Example #3

File: index.py Project: almightyGOSU/CZ4045-NLP-King-Chody-Almighty-Backend

def getDocumentCount():
	return str(CCorpusManager.getDocumentsCount())

Example #4

File: CSourceManager.py Project: almightyGOSU/CZ4045-NLP-King-Chody-Almighty-Backend

def getFullSource():

	for intCount in range(1, CCorpusManager.getDocumentsCount() + 1):
		yield "</br>" + CFileManager.readFromFile(str(intCount))