def getDocsTermsByClass(items): a = [] # items = readDocsByClass(v) for item in items: word_array = utils.extractWords(item[5] + ' ' + item[6]) a.extend(utils.arrayToLowerAndUnique(word_array)) return a
def getVocabulary(): a = [] items = readTrainingDataToArray() for item in items: word_array = utils.extractWords(item[5] + ' ' + item[6]) a.extend(utils.arrayToLowerAndUnique(word_array)) return a