In [75]:
#coding=utf-8
import xml.etree.ElementTree as ET
In [429]:
fileRoot = 'ES2015c.D.words'
In [430]:
fileName = '../AMI_corpus_resources/words/' + fileRoot + '.xml'
In [431]:
tree = ET.parse(fileName)
In [432]:
root = tree.getroot()
In [433]:
# load stop word list
stopWordFileName = '../../open_rogets_1.4/stops.txt'
stopFile = open(stopWordFileName, 'r+')
stopFile.seek(0)
# load stop word into array
stopWordList = []
for line in stopFile:
stopWordList.append(line.strip())
In [434]:
# check whether the given word is in the stop word list
# if does, return True
# else, return False
def checkStopWord(givenWord):
# check the word in the stop word list or not
exist = False
for stopWord in stopWordList:
if givenWord == stopWord:
exist = True
return exist
In [435]:
# load file content into dictionary
wordsInFile = dict()
for w in root.findall('w'):
wordDetail = {}
if w.text in wordsInFile:
# get current word detail
wordDetail = wordsInFile[w.text]
#update occurance times
times = wordDetail['occurance']
times = times + 1
wordDetail['occurance'] = times
# store the word detail
wordsInFile[w.text] = wordDetail
else:
wordDetail = {'occurance': 1, 'isStopWord':False}
wordsInFile[w.text] = wordDetail
In [436]:
print wordsInFile
{'all': {'occurance': 3, 'isStopWord': False}, 'fashion-watchers': {'occurance': 1, 'isStopWord': False}, 'pointing': {'occurance': 2, 'isStopWord': False}, 'channe': {'occurance': 1, 'isStopWord': False}, 'themes': {'occurance': 1, 'isStopWord': False}, 'skin': {'occurance': 1, 'isStopWord': False}, 'go': {'occurance': 3, 'isStopWord': False}, 'tw': {'occurance': 1, 'isStopWord': False}, 'to': {'occurance': 23, 'isStopWord': False}, 'under': {'occurance': 2, 'isStopWord': False}, 'Not': {'occurance': 1, 'isStopWord': False}, 'include': {'occurance': 1, 'isStopWord': False}, 'straightforward': {'occurance': 1, 'isStopWord': False}, 'very': {'occurance': 2, 'isStopWord': False}, "I'd": {'occurance': 1, 'isStopWord': False}, 'every': {'occurance': 2, 'isStopWord': False}, "we'll": {'occurance': 2, 'isStopWord': False}, 'Just': {'occurance': 1, 'isStopWord': False}, 'did': {'occurance': 1, 'isStopWord': False}, 'list': {'occurance': 1, 'isStopWord': False}, 'mm': {'occurance': 1, 'isStopWord': False}, 'clothes': {'occurance': 1, 'isStopWord': False}, 'talked': {'occurance': 2, 'isStopWord': False}, 'says': {'occurance': 1, 'isStopWord': False}, "you'd": {'occurance': 2, 'isStopWord': False}, 'ten': {'occurance': 1, 'isStopWord': False}, 'Should': {'occurance': 1, 'isStopWord': False}, 'second': {'occurance': 1, 'isStopWord': False}, 'design': {'occurance': 3, 'isStopWord': False}, 'web-based': {'occurance': 1, 'isStopWord': False}, 'what': {'occurance': 4, 'isStopWord': False}, 'preferences': {'occurance': 1, 'isStopWord': False}, 'current': {'occurance': 1, 'isStopWord': False}, 'revitalises': {'occurance': 1, 'isStopWord': False}, 'international': {'occurance': 1, 'isStopWord': False}, 'appeal': {'occurance': 1, 'isStopWord': False}, 'method': {'occurance': 1, 'isStopWord': False}, 'contrast': {'occurance': 1, 'isStopWord': False}, 'Mm': {'occurance': 2, 'isStopWord': False}, 'here': {'occurance': 2, 'isStopWord': False}, 'hours': {'occurance': 1, 'isStopWord': False}, 'Alright': {'occurance': 1, 'isStopWord': False}, "aren't": {'occurance': 1, 'isStopWord': False}, 'fifteen': {'occurance': 1, 'isStopWord': False}, 'casing': {'occurance': 1, 'isStopWord': False}, 'changed': {'occurance': 1, 'isStopWord': False}, 'reports': {'occurance': 1, 'isStopWord': False}, 'options': {'occurance': 1, 'isStopWord': False}, 'pitch': {'occurance': 1, 'isStopWord': False}, "It's": {'occurance': 1, 'isStopWord': False}, 'shoved': {'occurance': 1, 'isStopWord': False}, 'user-friendliness': {'occurance': 1, 'isStopWord': False}, 'technologically': {'occurance': 1, 'isStopWord': False}, 'use': {'occurance': 4, 'isStopWord': False}, 'from': {'occurance': 5, 'isStopWord': False}, 'spoke': {'occurance': 1, 'isStopWord': False}, 'would': {'occurance': 2, 'isStopWord': False}, 'ploy': {'occurance': 1, 'isStopWord': False}, 'two': {'occurance': 3, 'isStopWord': False}, 'live': {'occurance': 1, 'isStopWord': False}, 'markets': {'occurance': 1, 'isStopWord': False}, 'today': {'occurance': 1, 'isStopWord': False}, 'more': {'occurance': 2, 'isStopWord': False}, 'So': {'occurance': 16, 'isStopWord': False}, 'stick': {'occurance': 4, 'isStopWord': False}, 'tying': {'occurance': 1, 'isStopWord': False}, 'Now': {'occurance': 1, 'isStopWord': False}, 'animals': {'occurance': 1, 'isStopWord': False}, 'room': {'occurance': 1, 'isStopWord': False}, 'hour': {'occurance': 3, 'isStopWord': False}, 'f': {'occurance': 1, 'isStopWord': False}, 'this': {'occurance': 5, 'isStopWord': False}, 'work': {'occurance': 2, 'isStopWord': False}, 'cat': {'occurance': 1, 'isStopWord': False}, 'can': {'occurance': 3, 'isStopWord': False}, 'my': {'occurance': 4, 'isStopWord': False}, 'control': {'occurance': 4, 'isStopWord': False}, 'plate': {'occurance': 1, 'isStopWord': False}, "didn't": {'occurance': 1, 'isStopWord': False}, 'high': {'occurance': 1, 'isStopWord': False}, 'something': {'occurance': 3, 'isStopWord': False}, 'want': {'occurance': 2, 'isStopWord': False}, 'times': {'occurance': 2, 'isStopWord': False}, 'frontwards': {'occurance': 1, 'isStopWord': False}, 'needs': {'occurance': 1, 'isStopWord': False}, 'turn': {'occurance': 1, 'isStopWord': False}, 'how': {'occurance': 3, 'isStopWord': False}, 'instead': {'occurance': 1, 'isStopWord': False}, 'okay': {'occurance': 1, 'isStopWord': False}, 'after': {'occurance': 1, 'isStopWord': False}, 'lighting': {'occurance': 3, 'isStopWord': False}, 'coming': {'occurance': 2, 'isStopWord': False}, 'guys': {'occurance': 2, 'isStopWord': False}, 'a': {'occurance': 24, 'isStopWord': False}, 'remember': {'occurance': 2, 'isStopWord': False}, 'third': {'occurance': 1, 'isStopWord': False}, 'maybe': {'occurance': 3, 'isStopWord': False}, 'One': {'occurance': 1, 'isStopWord': False}, 'sh': {'occurance': 1, 'isStopWord': False}, 'so': {'occurance': 10, 'isStopWord': False}, "that's": {'occurance': 3, 'isStopWord': False}, 'shoes': {'occurance': 1, 'isStopWord': False}, "don't": {'occurance': 2, 'isStopWord': False}, 'essence': {'occurance': 1, 'isStopWord': False}, 'over': {'occurance': 1, 'isStopWord': False}, 'mainly': {'occurance': 1, 'isStopWord': False}, 'kickoff': {'occurance': 1, 'isStopWord': False}, 'years': {'occurance': 1, 'isStopWord': False}, 'course': {'occurance': 1, 'isStopWord': False}, 'looks': {'occurance': 1, 'isStopWord': False}, 'innovative': {'occurance': 1, 'isStopWord': False}, 'before': {'occurance': 1, 'isStopWord': False}, 'style': {'occurance': 2, 'isStopWord': False}, 'personal': {'occurance': 2, 'isStopWord': False}, ',': {'occurance': 98, 'isStopWord': False}, 'actually': {'occurance': 1, 'isStopWord': False}, 'might': {'occurance': 3, 'isStopWord': False}, 'versus': {'occurance': 1, 'isStopWord': False}, 'then': {'occurance': 2, 'isStopWord': False}, 'them': {'occurance': 2, 'isStopWord': False}, 'good': {'occurance': 2, 'isStopWord': False}, 'material': {'occurance': 1, 'isStopWord': False}, 'Because': {'occurance': 1, 'isStopWord': False}, 'they': {'occurance': 4, 'isStopWord': False}, 'not': {'occurance': 2, 'isStopWord': False}, 'brightness': {'occurance': 1, 'isStopWord': False}, 'Milan': {'occurance': 2, 'isStopWord': False}, 'each': {'occurance': 1, 'isStopWord': False}, 'mean': {'occurance': 4, 'isStopWord': False}, 'schemes': {'occurance': 1, 'isStopWord': False}, 'doing': {'occurance': 1, 'isStopWord': False}, 'house': {'occurance': 1, 'isStopWord': False}, 'hard': {'occurance': 2, 'isStopWord': False}, 'yeah': {'occurance': 3, 'isStopWord': False}, "we'd": {'occurance': 1, 'isStopWord': False}, 'year': {'occurance': 2, 'isStopWord': False}, 'our': {'occurance': 5, 'isStopWord': False}, 'out': {'occurance': 1, 'isStopWord': False}, 'living': {'occurance': 1, 'isStopWord': False}, 'research': {'occurance': 2, 'isStopWord': False}, 'shows': {'occurance': 1, 'isStopWord': False}, "doesn't": {'occurance': 2, 'isStopWord': False}, 'This': {'occurance': 1, 'isStopWord': False}, 'guess': {'occurance': 5, 'isStopWord': False}, 'wanted': {'occurance': 1, 'isStopWord': False}, 'scary': {'occurance': 1, 'isStopWord': False}, 'dunno': {'occurance': 2, 'isStopWord': False}, 'could': {'occurance': 3, 'isStopWord': False}, 'keep': {'occurance': 2, 'isStopWord': False}, 'thing': {'occurance': 2, 'isStopWord': False}, 'w': {'occurance': 1, 'isStopWord': False}, 'think': {'occurance': 6, 'isStopWord': False}, 'first': {'occurance': 2, 'isStopWord': False}, 'already': {'occurance': 2, 'isStopWord': False}, 'saving': {'occurance': 1, 'isStopWord': False}, 'feel': {'occurance': 1, 'isStopWord': False}, 'There': {'occurance': 2, 'isStopWord': False}, 'fancy': {'occurance': 1, 'isStopWord': False}, 'one': {'occurance': 9, 'isStopWord': False}, 'done': {'occurance': 2, 'isStopWord': False}, 'sounds': {'occurance': 2, 'isStopWord': False}, "you're": {'occurance': 2, 'isStopWord': False}, 'little': {'occurance': 1, 'isStopWord': False}, 'Okay': {'occurance': 15, 'isStopWord': False}, 'their': {'occurance': 1, 'isStopWord': False}, 'twenty': {'occurance': 1, 'isStopWord': False}, 'Screen': {'occurance': 1, 'isStopWord': False}, 'plastic': {'occurance': 2, 'isStopWord': False}, 'priority': {'occurance': 1, 'isStopWord': False}, 'contra': {'occurance': 1, 'isStopWord': False}, 'too': {'occurance': 1, 'isStopWord': False}, 'which': {'occurance': 1, 'isStopWord': False}, 'Remote': {'occurance': 1, 'isStopWord': False}, 'option': {'occurance': 1, 'isStopWord': False}, 'that': {'occurance': 27, 'isStopWord': False}, 'took': {'occurance': 1, 'isStopWord': False}, 'light-up': {'occurance': 1, 'isStopWord': False}, 'somewhat': {'occurance': 1, 'isStopWord': False}, 'believe': {'occurance': 1, 'isStopWord': False}, 'than': {'occurance': 1, 'isStopWord': False}, 'television': {'occurance': 1, 'isStopWord': False}, 'zero': {'occurance': 2, 'isStopWord': False}, 'animal-based': {'occurance': 1, 'isStopWord': False}, 'fruits': {'occurance': 1, 'isStopWord': False}, 'and': {'occurance': 22, 'isStopWord': False}, 'fashion': {'occurance': 2, 'isStopWord': False}, 'mind': {'occurance': 1, 'isStopWord': False}, 'mine': {'occurance': 1, 'isStopWord': False}, 'talking': {'occurance': 1, 'isStopWord': False}, 'say': {'occurance': 2, 'isStopWord': False}, 'have': {'occurance': 13, 'isStopWord': False}, 'need': {'occurance': 1, 'isStopWord': False}, 'seen': {'occurance': 1, 'isStopWord': False}, 'any': {'occurance': 1, 'isStopWord': False}, 'Do': {'occurance': 1, 'isStopWord': False}, 'depending': {'occurance': 1, 'isStopWord': False}, 'able': {'occurance': 1, 'isStopWord': False}, 'ideas': {'occurance': 1, 'isStopWord': False}, 'also': {'occurance': 2, 'isStopWord': False}, 'They': {'occurance': 1, 'isStopWord': False}, 'channel': {'occurance': 1, 'isStopWord': False}, 'though': {'occurance': 1, 'isStopWord': False}, 'track': {'occurance': 1, 'isStopWord': False}, 'most': {'occurance': 5, 'isStopWord': False}, 'regular': {'occurance': 1, 'isStopWord': False}, 'eight': {'occurance': 1, 'isStopWord': False}, "'cause": {'occurance': 2, 'isStopWord': False}, 'device': {'occurance': 1, 'isStopWord': False}, 'The': {'occurance': 8, 'isStopWord': False}, 'trend-watching': {'occurance': 1, 'isStopWord': False}, 'stereo': {'occurance': 1, 'isStopWord': False}, 'considered': {'occurance': 2, 'isStopWord': False}, 'sometimes': {'occurance': 2, 'isStopWord': False}, 'Some': {'occurance': 1, 'isStopWord': False}, 'dog': {'occurance': 1, 'isStopWord': False}, 'looked': {'occurance': 1, 'isStopWord': False}, 'fashions': {'occurance': 1, 'isStopWord': False}, 'If': {'occurance': 1, 'isStopWord': False}, 'annoyed': {'occurance': 1, 'isStopWord': False}, 'saying': {'occurance': 2, 'isStopWord': False}, 'Paris': {'occurance': 2, 'isStopWord': False}, 'infrared': {'occurance': 1, 'isStopWord': False}, 'bring': {'occurance': 1, 'isStopWord': False}, 'based': {'occurance': 1, 'isStopWord': False}, 'with': {'occurance': 10, 'isStopWord': False}, 'controls': {'occurance': 1, 'isStopWord': False}, 'should': {'occurance': 5, 'isStopWord': False}, 'buttons': {'occurance': 6, 'isStopWord': False}, 'theme': {'occurance': 1, 'isStopWord': False}, 'going': {'occurance': 1, 'isStopWord': False}, 'black': {'occurance': 1, 'isStopWord': False}, 'York': {'occurance': 1, 'isStopWord': False}, 'pretty': {'occurance': 1, 'isStopWord': False}, 'do': {'occurance': 3, 'isStopWord': False}, 'get': {'occurance': 1, 'isStopWord': False}, 'kind': {'occurance': 2, 'isStopWord': False}, 'T_V_s': {'occurance': 1, 'isStopWord': False}, 'remotes': {'occurance': 2, 'isStopWord': False}, 'cushion': {'occurance': 1, 'isStopWord': False}, 'report': {'occurance': 2, 'isStopWord': False}, 'Right': {'occurance': 4, 'isStopWord': False}, 'settings': {'occurance': 2, 'isStopWord': False}, 'twice': {'occurance': 1, 'isStopWord': False}, 'questions': {'occurance': 1, 'isStopWord': False}, 'shake': {'occurance': 1, 'isStopWord': False}, "hasn't": {'occurance': 1, 'isStopWord': False}, 'Mm-hmm': {'occurance': 13, 'isStopWord': False}, 'see': {'occurance': 3, 'isStopWord': False}, 'are': {'occurance': 3, 'isStopWord': False}, 'That': {'occurance': 1, 'isStopWord': False}, 'movie': {'occurance': 1, 'isStopWord': False}, 'purses': {'occurance': 1, 'isStopWord': False}, 'colours': {'occurance': 1, 'isStopWord': False}, "there's": {'occurance': 1, 'isStopWord': False}, 'across': {'occurance': 1, 'isStopWord': False}, 'pictured': {'occurance': 1, 'isStopWord': False}, 'we': {'occurance': 13, 'isStopWord': False}, 'interface': {'occurance': 2, 'isStopWord': False}, 'come': {'occurance': 2, 'isStopWord': False}, 'kitchen': {'occurance': 1, 'isStopWord': False}, 'c': {'occurance': 1, 'isStopWord': False}, 'last': {'occurance': 1, 'isStopWord': False}, 'many': {'occurance': 1, 'isStopWord': False}, 'according': {'occurance': 1, 'isStopWord': False}, 'presentations': {'occurance': 2, 'isStopWord': False}, 'Yes': {'occurance': 1, 'isStopWord': False}, 'comes': {'occurance': 2, 'isStopWord': False}, 'ch': {'occurance': 1, 'isStopWord': False}, 'tone': {'occurance': 1, 'isStopWord': False}, 'clashing': {'occurance': 1, 'isStopWord': False}, 'point': {'occurance': 4, 'isStopWord': False}, 'battery': {'occurance': 3, 'isStopWord': False}, 'Sorry': {'occurance': 1, 'isStopWord': False}, 'olives': {'occurance': 1, 'isStopWord': False}, 'light': {'occurance': 2, 'isStopWord': False}, "I've": {'occurance': 2, 'isStopWord': False}, "it's": {'occurance': 9, 'isStopWord': False}, 'three': {'occurance': 1, 'isStopWord': False}, 'But': {'occurance': 3, 'isStopWord': False}, '.': {'occurance': 128, 'isStopWord': False}, 'basic': {'occurance': 1, 'isStopWord': False}, 'expected': {'occurance': 1, 'isStopWord': False}, 'meeting': {'occurance': 2, 'isStopWord': False}, 'graphical': {'occurance': 1, 'isStopWord': False}, 'easy': {'occurance': 2, 'isStopWord': False}, "That's": {'occurance': 2, 'isStopWord': False}, 'And': {'occurance': 6, 'isStopWord': False}, 'has': {'occurance': 1, 'isStopWord': False}, 'look': {'occurance': 1, 'isStopWord': False}, 'these': {'occurance': 1, 'isStopWord': False}, 'tastes': {'occurance': 1, 'isStopWord': False}, 'will': {'occurance': 3, 'isStopWord': False}, 'while': {'occurance': 1, 'isStopWord': False}, 'fun': {'occurance': 1, 'isStopWord': False}, 'PowerPoint': {'occurance': 2, 'isStopWord': False}, 'picturing': {'occurance': 1, 'isStopWord': False}, 'pack': {'occurance': 1, 'isStopWord': False}, 'almost': {'occurance': 1, 'isStopWord': False}, 'is': {'occurance': 14, 'isStopWord': False}, 'it': {'occurance': 23, 'isStopWord': False}, 'experts': {'occurance': 1, 'isStopWord': False}, 'Pretty': {'occurance': 1, 'isStopWord': False}, 'in': {'occurance': 8, 'isStopWord': False}, 'You': {'occurance': 1, 'isStopWord': False}, 'casings': {'occurance': 1, 'isStopWord': False}, 'if': {'occurance': 10, 'isStopWord': False}, 'different': {'occurance': 2, 'isStopWord': False}, 'shorter': {'occurance': 1, 'isStopWord': False}, 'vegetable': {'occurance': 1, 'isStopWord': False}, 'About': {'occurance': 1, 'isStopWord': False}, 'used': {'occurance': 3, 'isStopWord': False}, 'temporary': {'occurance': 1, 'isStopWord': False}, 'I': {'occurance': 27, 'isStopWord': False}, 'Maybe': {'occurance': 1, 'isStopWord': False}, 'running': {'occurance': 1, 'isStopWord': False}, 'fruit': {'occurance': 1, 'isStopWord': False}, 'purpose': {'occurance': 1, 'isStopWord': False}, 'recent': {'occurance': 1, 'isStopWord': False}, 'off': {'occurance': 2, 'isStopWord': False}, 'i': {'occurance': 1, 'isStopWord': False}, 'colour': {'occurance': 3, 'isStopWord': False}, 'fashionable': {'occurance': 1, 'isStopWord': False}, 'well': {'occurance': 2, 'isStopWord': False}, 'It': {'occurance': 2, 'isStopWord': False}, 'obviously': {'occurance': 1, 'isStopWord': False}, 'command': {'occurance': 1, 'isStopWord': False}, 'Have': {'occurance': 1, 'isStopWord': False}, 'y': {'occurance': 1, 'isStopWord': False}, 'the': {'occurance': 37, 'isStopWord': False}, 'audio': {'occurance': 1, 'isStopWord': False}, 'know': {'occurance': 8, 'isStopWord': False}, 'just': {'occurance': 7, 'isStopWord': False}, 'true': {'occurance': 1, 'isStopWord': False}, 'batteries': {'occurance': 1, 'isStopWord': False}, 'aspect': {'occurance': 3, 'isStopWord': False}, 'olds': {'occurance': 1, 'isStopWord': False}, 'previous': {'occurance': 1, 'isStopWord': False}, 'web': {'occurance': 1, 'isStopWord': False}, 'thinking': {'occurance': 1, 'isStopWord': False}, "we've": {'occurance': 4, 'isStopWord': False}, 'seems': {'occurance': 1, 'isStopWord': False}, 'simple': {'occurance': 1, 'isStopWord': False}, 'Although': {'occurance': 1, 'isStopWord': False}, 'relevance': {'occurance': 2, 'isStopWord': False}, 'look-and-feel': {'occurance': 2, 'isStopWord': False}, 'aspects': {'occurance': 1, 'isStopWord': False}, 'around': {'occurance': 1, 'isStopWord': False}, 'Oh': {'occurance': 6, 'isStopWord': False}, 'Mono': {'occurance': 1, 'isStopWord': False}, 'gonna': {'occurance': 2, 'isStopWord': False}, 'five': {'occurance': 4, 'isStopWord': False}, 'T': {'occurance': 1, 'isStopWord': False}, 'bit': {'occurance': 2, 'isStopWord': False}, 'Or': {'occurance': 3, 'isStopWord': False}, 'furniture': {'occurance': 1, 'isStopWord': False}, 'insert': {'occurance': 1, 'isStopWord': False}, 'like': {'occurance': 5, 'isStopWord': False}, 'success': {'occurance': 1, 'isStopWord': False}, 'specific': {'occurance': 1, 'isStopWord': False}, 't': {'occurance': 1, 'isStopWord': False}, 'Again': {'occurance': 1, 'isStopWord': False}, 'because': {'occurance': 2, 'isStopWord': False}, 'often': {'occurance': 1, 'isStopWord': False}, 'people': {'occurance': 2, 'isStopWord': False}, 'some': {'occurance': 3, 'isStopWord': False}, 'back': {'occurance': 2, 'isStopWord': False}, 'trendy': {'occurance': 1, 'isStopWord': False}, 'New': {'occurance': 1, 'isStopWord': False}, 'scale': {'occurance': 1, 'isStopWord': False}, 'for': {'occurance': 6, 'isStopWord': False}, 'affects': {'occurance': 1, 'isStopWord': False}, 'face-plates': {'occurance': 3, 'isStopWord': False}, '?': {'occurance': 12, 'isStopWord': False}, 'Something': {'occurance': 1, 'isStopWord': False}, 'be': {'occurance': 11, 'isStopWord': False}, 'power': {'occurance': 1, 'isStopWord': False}, 'marketing': {'occurance': 1, 'isStopWord': False}, 'Um': {'occurance': 15, 'isStopWord': False}, 'Command': {'occurance': 1, 'isStopWord': False}, 'Uh': {'occurance': 2, 'isStopWord': False}, 'Think': {'occurance': 1, 'isStopWord': False}, 'on': {'occurance': 10, 'isStopWord': False}, 'about': {'occurance': 4, 'isStopWord': False}, 'actual': {'occurance': 1, 'isStopWord': False}, 'working': {'occurance': 1, 'isStopWord': False}, 'getting': {'occurance': 1, 'isStopWord': False}, 'of': {'occurance': 22, 'isStopWord': False}, 'discussed': {'occurance': 1, 'isStopWord': False}, 'draf': {'occurance': 1, 'isStopWord': False}, 'wanna': {'occurance': 1, 'isStopWord': False}, 'or': {'occurance': 3, 'isStopWord': False}, 'functional': {'occurance': 1, 'isStopWord': False}, 'right': {'occurance': 2, 'isStopWord': False}, 'been': {'occurance': 3, 'isStopWord': False}, 'your': {'occurance': 4, 'isStopWord': False}, 'her': {'occurance': 1, 'isStopWord': False}, 'vegetables': {'occurance': 1, 'isStopWord': False}, 'there': {'occurance': 4, 'isStopWord': False}, "we're": {'occurance': 3, 'isStopWord': False}, 'lot': {'occurance': 3, 'isStopWord': False}, 'was': {'occurance': 9, 'isStopWord': False}, 'spongy': {'occurance': 1, 'isStopWord': False}, 'Well': {'occurance': 2, 'isStopWord': False}, 'but': {'occurance': 2, 'isStopWord': False}, 'idea': {'occurance': 2, 'isStopWord': False}, 'volume': {'occurance': 1, 'isStopWord': False}, 'granny': {'occurance': 1, 'isStopWord': False}, 'kitchenware': {'occurance': 1, 'isStopWord': False}, "I'll": {'occurance': 1, 'isStopWord': False}, "they're": {'occurance': 1, 'isStopWord': False}, 'up': {'occurance': 5, 'isStopWord': False}, 'us': {'occurance': 1, 'isStopWord': False}, 'um': {'occurance': 6, 'isStopWord': False}, 'strawberry': {'occurance': 1, 'isStopWord': False}, 'uh': {'occurance': 7, 'isStopWord': False}, 'called': {'occurance': 1, 'isStopWord': False}, 'sales': {'occurance': 1, 'isStopWord': False}, 'an': {'occurance': 4, 'isStopWord': False}, 'as': {'occurance': 6, 'isStopWord': False}, 'at': {'occurance': 2, 'isStopWord': False}, 'Real': {'occurance': 1, 'isStopWord': False}, 'ideally': {'occurance': 1, 'isStopWord': False}, 'when': {'occurance': 1, 'isStopWord': False}, 'rubber': {'occurance': 3, 'isStopWord': False}, 'other': {'occurance': 1, 'isStopWord': False}, 'you': {'occurance': 19, 'isStopWord': False}, 'Like': {'occurance': 3, 'isStopWord': False}, 'favourite': {'occurance': 1, 'isStopWord': False}, 'update': {'occurance': 1, 'isStopWord': False}, 'important': {'occurance': 5, 'isStopWord': False}, 'kinda': {'occurance': 1, 'isStopWord': False}, 'sides': {'occurance': 1, 'isStopWord': False}, 'remote': {'occurance': 7, 'isStopWord': False}, 'T_V_': {'occurance': 2, 'isStopWord': False}, 'age': {'occurance': 1, 'isStopWord': False}, 'Yeah': {'occurance': 13, 'isStopWord': False}, 'time': {'occurance': 1, 'isStopWord': False}, 'everywhere': {'occurance': 1, 'isStopWord': False}, 'once': {'occurance': 2, 'isStopWord': False}, 'having': {'occurance': 1, 'isStopWord': False}, 'basically': {'occurance': 1, 'isStopWord': False}}
In [437]:
# check if the word is stop word or not
for key, wordDetail in wordsInFile.iteritems():
exist = checkStopWord(key.lower())
wordDetail['isStopWord'] = exist
# store the word detail
wordsInFile[key] = wordDetail
In [438]:
print wordsInFile
{'all': {'occurance': 3, 'isStopWord': True}, 'fashion-watchers': {'occurance': 1, 'isStopWord': False}, 'pointing': {'occurance': 2, 'isStopWord': True}, 'channe': {'occurance': 1, 'isStopWord': False}, 'themes': {'occurance': 1, 'isStopWord': False}, 'skin': {'occurance': 1, 'isStopWord': False}, 'go': {'occurance': 3, 'isStopWord': True}, 'tw': {'occurance': 1, 'isStopWord': False}, 'to': {'occurance': 23, 'isStopWord': True}, 'under': {'occurance': 2, 'isStopWord': True}, 'Not': {'occurance': 1, 'isStopWord': True}, 'include': {'occurance': 1, 'isStopWord': False}, 'straightforward': {'occurance': 1, 'isStopWord': False}, 'very': {'occurance': 2, 'isStopWord': True}, "I'd": {'occurance': 1, 'isStopWord': True}, 'every': {'occurance': 2, 'isStopWord': True}, "we'll": {'occurance': 2, 'isStopWord': True}, 'Just': {'occurance': 1, 'isStopWord': True}, 'did': {'occurance': 1, 'isStopWord': True}, 'list': {'occurance': 1, 'isStopWord': False}, 'mm': {'occurance': 1, 'isStopWord': True}, 'clothes': {'occurance': 1, 'isStopWord': False}, 'talked': {'occurance': 2, 'isStopWord': False}, 'says': {'occurance': 1, 'isStopWord': True}, "you'd": {'occurance': 2, 'isStopWord': True}, 'ten': {'occurance': 1, 'isStopWord': True}, 'Should': {'occurance': 1, 'isStopWord': True}, 'second': {'occurance': 1, 'isStopWord': True}, 'design': {'occurance': 3, 'isStopWord': False}, 'web-based': {'occurance': 1, 'isStopWord': False}, 'what': {'occurance': 4, 'isStopWord': True}, 'preferences': {'occurance': 1, 'isStopWord': False}, 'current': {'occurance': 1, 'isStopWord': False}, 'revitalises': {'occurance': 1, 'isStopWord': False}, 'international': {'occurance': 1, 'isStopWord': False}, 'appeal': {'occurance': 1, 'isStopWord': False}, 'method': {'occurance': 1, 'isStopWord': False}, 'contrast': {'occurance': 1, 'isStopWord': False}, 'Mm': {'occurance': 2, 'isStopWord': True}, 'here': {'occurance': 2, 'isStopWord': True}, 'hours': {'occurance': 1, 'isStopWord': False}, 'Alright': {'occurance': 1, 'isStopWord': False}, "aren't": {'occurance': 1, 'isStopWord': True}, 'fifteen': {'occurance': 1, 'isStopWord': True}, 'casing': {'occurance': 1, 'isStopWord': False}, 'changed': {'occurance': 1, 'isStopWord': False}, 'reports': {'occurance': 1, 'isStopWord': False}, 'options': {'occurance': 1, 'isStopWord': False}, 'pitch': {'occurance': 1, 'isStopWord': False}, "It's": {'occurance': 1, 'isStopWord': True}, 'shoved': {'occurance': 1, 'isStopWord': False}, 'user-friendliness': {'occurance': 1, 'isStopWord': False}, 'technologically': {'occurance': 1, 'isStopWord': False}, 'use': {'occurance': 4, 'isStopWord': True}, 'from': {'occurance': 5, 'isStopWord': True}, 'spoke': {'occurance': 1, 'isStopWord': False}, 'would': {'occurance': 2, 'isStopWord': True}, 'ploy': {'occurance': 1, 'isStopWord': False}, 'two': {'occurance': 3, 'isStopWord': True}, 'live': {'occurance': 1, 'isStopWord': False}, 'markets': {'occurance': 1, 'isStopWord': False}, 'today': {'occurance': 1, 'isStopWord': True}, 'more': {'occurance': 2, 'isStopWord': True}, 'So': {'occurance': 16, 'isStopWord': True}, 'stick': {'occurance': 4, 'isStopWord': False}, 'tying': {'occurance': 1, 'isStopWord': False}, 'Now': {'occurance': 1, 'isStopWord': True}, 'animals': {'occurance': 1, 'isStopWord': False}, 'room': {'occurance': 1, 'isStopWord': True}, 'hour': {'occurance': 3, 'isStopWord': False}, 'f': {'occurance': 1, 'isStopWord': True}, 'this': {'occurance': 5, 'isStopWord': True}, 'work': {'occurance': 2, 'isStopWord': True}, 'cat': {'occurance': 1, 'isStopWord': False}, 'can': {'occurance': 3, 'isStopWord': True}, 'my': {'occurance': 4, 'isStopWord': True}, 'control': {'occurance': 4, 'isStopWord': False}, 'plate': {'occurance': 1, 'isStopWord': False}, "didn't": {'occurance': 1, 'isStopWord': True}, 'high': {'occurance': 1, 'isStopWord': True}, 'something': {'occurance': 3, 'isStopWord': True}, 'want': {'occurance': 2, 'isStopWord': True}, 'times': {'occurance': 2, 'isStopWord': False}, 'frontwards': {'occurance': 1, 'isStopWord': False}, 'needs': {'occurance': 1, 'isStopWord': True}, 'turn': {'occurance': 1, 'isStopWord': True}, 'how': {'occurance': 3, 'isStopWord': True}, 'instead': {'occurance': 1, 'isStopWord': True}, 'okay': {'occurance': 1, 'isStopWord': True}, 'after': {'occurance': 1, 'isStopWord': True}, 'lighting': {'occurance': 3, 'isStopWord': False}, 'coming': {'occurance': 2, 'isStopWord': False}, 'guys': {'occurance': 2, 'isStopWord': False}, 'a': {'occurance': 24, 'isStopWord': True}, 'remember': {'occurance': 2, 'isStopWord': False}, 'third': {'occurance': 1, 'isStopWord': True}, 'maybe': {'occurance': 3, 'isStopWord': True}, 'One': {'occurance': 1, 'isStopWord': True}, 'sh': {'occurance': 1, 'isStopWord': False}, 'so': {'occurance': 10, 'isStopWord': True}, "that's": {'occurance': 3, 'isStopWord': True}, 'shoes': {'occurance': 1, 'isStopWord': False}, "don't": {'occurance': 2, 'isStopWord': True}, 'essence': {'occurance': 1, 'isStopWord': False}, 'over': {'occurance': 1, 'isStopWord': True}, 'mainly': {'occurance': 1, 'isStopWord': True}, 'kickoff': {'occurance': 1, 'isStopWord': False}, 'years': {'occurance': 1, 'isStopWord': True}, 'course': {'occurance': 1, 'isStopWord': True}, 'looks': {'occurance': 1, 'isStopWord': True}, 'innovative': {'occurance': 1, 'isStopWord': False}, 'before': {'occurance': 1, 'isStopWord': True}, 'style': {'occurance': 2, 'isStopWord': False}, 'personal': {'occurance': 2, 'isStopWord': False}, ',': {'occurance': 98, 'isStopWord': False}, 'actually': {'occurance': 1, 'isStopWord': True}, 'might': {'occurance': 3, 'isStopWord': True}, 'versus': {'occurance': 1, 'isStopWord': False}, 'then': {'occurance': 2, 'isStopWord': True}, 'them': {'occurance': 2, 'isStopWord': True}, 'good': {'occurance': 2, 'isStopWord': True}, 'material': {'occurance': 1, 'isStopWord': False}, 'Because': {'occurance': 1, 'isStopWord': True}, 'they': {'occurance': 4, 'isStopWord': True}, 'not': {'occurance': 2, 'isStopWord': True}, 'brightness': {'occurance': 1, 'isStopWord': False}, 'Milan': {'occurance': 2, 'isStopWord': False}, 'each': {'occurance': 1, 'isStopWord': True}, 'mean': {'occurance': 4, 'isStopWord': True}, 'schemes': {'occurance': 1, 'isStopWord': False}, 'doing': {'occurance': 1, 'isStopWord': True}, 'house': {'occurance': 1, 'isStopWord': False}, 'hard': {'occurance': 2, 'isStopWord': False}, 'yeah': {'occurance': 3, 'isStopWord': False}, "we'd": {'occurance': 1, 'isStopWord': True}, 'year': {'occurance': 2, 'isStopWord': True}, 'our': {'occurance': 5, 'isStopWord': True}, 'out': {'occurance': 1, 'isStopWord': True}, 'living': {'occurance': 1, 'isStopWord': False}, 'research': {'occurance': 2, 'isStopWord': False}, 'shows': {'occurance': 1, 'isStopWord': True}, "doesn't": {'occurance': 2, 'isStopWord': True}, 'This': {'occurance': 1, 'isStopWord': True}, 'guess': {'occurance': 5, 'isStopWord': False}, 'wanted': {'occurance': 1, 'isStopWord': True}, 'scary': {'occurance': 1, 'isStopWord': False}, 'dunno': {'occurance': 2, 'isStopWord': False}, 'could': {'occurance': 3, 'isStopWord': True}, 'keep': {'occurance': 2, 'isStopWord': True}, 'thing': {'occurance': 2, 'isStopWord': True}, 'w': {'occurance': 1, 'isStopWord': True}, 'think': {'occurance': 6, 'isStopWord': True}, 'first': {'occurance': 2, 'isStopWord': True}, 'already': {'occurance': 2, 'isStopWord': True}, 'saving': {'occurance': 1, 'isStopWord': False}, 'feel': {'occurance': 1, 'isStopWord': False}, 'There': {'occurance': 2, 'isStopWord': True}, 'fancy': {'occurance': 1, 'isStopWord': False}, 'one': {'occurance': 9, 'isStopWord': True}, 'done': {'occurance': 2, 'isStopWord': True}, 'sounds': {'occurance': 2, 'isStopWord': False}, "you're": {'occurance': 2, 'isStopWord': True}, 'little': {'occurance': 1, 'isStopWord': True}, 'Okay': {'occurance': 15, 'isStopWord': True}, 'their': {'occurance': 1, 'isStopWord': True}, 'twenty': {'occurance': 1, 'isStopWord': True}, 'Screen': {'occurance': 1, 'isStopWord': False}, 'plastic': {'occurance': 2, 'isStopWord': False}, 'priority': {'occurance': 1, 'isStopWord': False}, 'contra': {'occurance': 1, 'isStopWord': False}, 'too': {'occurance': 1, 'isStopWord': True}, 'which': {'occurance': 1, 'isStopWord': True}, 'Remote': {'occurance': 1, 'isStopWord': False}, 'option': {'occurance': 1, 'isStopWord': False}, 'that': {'occurance': 27, 'isStopWord': True}, 'took': {'occurance': 1, 'isStopWord': True}, 'light-up': {'occurance': 1, 'isStopWord': False}, 'somewhat': {'occurance': 1, 'isStopWord': True}, 'believe': {'occurance': 1, 'isStopWord': True}, 'than': {'occurance': 1, 'isStopWord': True}, 'television': {'occurance': 1, 'isStopWord': False}, 'zero': {'occurance': 2, 'isStopWord': True}, 'animal-based': {'occurance': 1, 'isStopWord': False}, 'fruits': {'occurance': 1, 'isStopWord': False}, 'and': {'occurance': 22, 'isStopWord': True}, 'fashion': {'occurance': 2, 'isStopWord': False}, 'mind': {'occurance': 1, 'isStopWord': False}, 'mine': {'occurance': 1, 'isStopWord': True}, 'talking': {'occurance': 1, 'isStopWord': False}, 'say': {'occurance': 2, 'isStopWord': True}, 'have': {'occurance': 13, 'isStopWord': True}, 'need': {'occurance': 1, 'isStopWord': True}, 'seen': {'occurance': 1, 'isStopWord': True}, 'any': {'occurance': 1, 'isStopWord': True}, 'Do': {'occurance': 1, 'isStopWord': True}, 'depending': {'occurance': 1, 'isStopWord': False}, 'able': {'occurance': 1, 'isStopWord': True}, 'ideas': {'occurance': 1, 'isStopWord': False}, 'also': {'occurance': 2, 'isStopWord': True}, 'They': {'occurance': 1, 'isStopWord': True}, 'channel': {'occurance': 1, 'isStopWord': False}, 'though': {'occurance': 1, 'isStopWord': True}, 'track': {'occurance': 1, 'isStopWord': False}, 'most': {'occurance': 5, 'isStopWord': True}, 'regular': {'occurance': 1, 'isStopWord': False}, 'eight': {'occurance': 1, 'isStopWord': True}, "'cause": {'occurance': 2, 'isStopWord': False}, 'device': {'occurance': 1, 'isStopWord': False}, 'The': {'occurance': 8, 'isStopWord': True}, 'trend-watching': {'occurance': 1, 'isStopWord': False}, 'stereo': {'occurance': 1, 'isStopWord': False}, 'considered': {'occurance': 2, 'isStopWord': False}, 'sometimes': {'occurance': 2, 'isStopWord': True}, 'Some': {'occurance': 1, 'isStopWord': True}, 'dog': {'occurance': 1, 'isStopWord': False}, 'looked': {'occurance': 1, 'isStopWord': False}, 'fashions': {'occurance': 1, 'isStopWord': False}, 'If': {'occurance': 1, 'isStopWord': True}, 'annoyed': {'occurance': 1, 'isStopWord': False}, 'saying': {'occurance': 2, 'isStopWord': True}, 'Paris': {'occurance': 2, 'isStopWord': False}, 'infrared': {'occurance': 1, 'isStopWord': False}, 'bring': {'occurance': 1, 'isStopWord': False}, 'based': {'occurance': 1, 'isStopWord': False}, 'with': {'occurance': 10, 'isStopWord': True}, 'controls': {'occurance': 1, 'isStopWord': False}, 'should': {'occurance': 5, 'isStopWord': True}, 'buttons': {'occurance': 6, 'isStopWord': False}, 'theme': {'occurance': 1, 'isStopWord': False}, 'going': {'occurance': 1, 'isStopWord': True}, 'black': {'occurance': 1, 'isStopWord': False}, 'York': {'occurance': 1, 'isStopWord': False}, 'pretty': {'occurance': 1, 'isStopWord': False}, 'do': {'occurance': 3, 'isStopWord': True}, 'get': {'occurance': 1, 'isStopWord': True}, 'kind': {'occurance': 2, 'isStopWord': False}, 'T_V_s': {'occurance': 1, 'isStopWord': False}, 'remotes': {'occurance': 2, 'isStopWord': False}, 'cushion': {'occurance': 1, 'isStopWord': False}, 'report': {'occurance': 2, 'isStopWord': False}, 'Right': {'occurance': 4, 'isStopWord': True}, 'settings': {'occurance': 2, 'isStopWord': False}, 'twice': {'occurance': 1, 'isStopWord': True}, 'questions': {'occurance': 1, 'isStopWord': False}, 'shake': {'occurance': 1, 'isStopWord': False}, "hasn't": {'occurance': 1, 'isStopWord': True}, 'Mm-hmm': {'occurance': 13, 'isStopWord': False}, 'see': {'occurance': 3, 'isStopWord': True}, 'are': {'occurance': 3, 'isStopWord': True}, 'That': {'occurance': 1, 'isStopWord': True}, 'movie': {'occurance': 1, 'isStopWord': False}, 'purses': {'occurance': 1, 'isStopWord': False}, 'colours': {'occurance': 1, 'isStopWord': False}, "there's": {'occurance': 1, 'isStopWord': True}, 'across': {'occurance': 1, 'isStopWord': True}, 'pictured': {'occurance': 1, 'isStopWord': False}, 'we': {'occurance': 13, 'isStopWord': True}, 'interface': {'occurance': 2, 'isStopWord': False}, 'come': {'occurance': 2, 'isStopWord': True}, 'kitchen': {'occurance': 1, 'isStopWord': False}, 'c': {'occurance': 1, 'isStopWord': True}, 'last': {'occurance': 1, 'isStopWord': True}, 'many': {'occurance': 1, 'isStopWord': True}, 'according': {'occurance': 1, 'isStopWord': True}, 'presentations': {'occurance': 2, 'isStopWord': False}, 'Yes': {'occurance': 1, 'isStopWord': True}, 'comes': {'occurance': 2, 'isStopWord': True}, 'ch': {'occurance': 1, 'isStopWord': False}, 'tone': {'occurance': 1, 'isStopWord': False}, 'clashing': {'occurance': 1, 'isStopWord': False}, 'point': {'occurance': 4, 'isStopWord': True}, 'battery': {'occurance': 3, 'isStopWord': False}, 'Sorry': {'occurance': 1, 'isStopWord': True}, 'olives': {'occurance': 1, 'isStopWord': False}, 'light': {'occurance': 2, 'isStopWord': False}, "I've": {'occurance': 2, 'isStopWord': True}, "it's": {'occurance': 9, 'isStopWord': True}, 'three': {'occurance': 1, 'isStopWord': True}, 'But': {'occurance': 3, 'isStopWord': True}, '.': {'occurance': 128, 'isStopWord': False}, 'basic': {'occurance': 1, 'isStopWord': False}, 'expected': {'occurance': 1, 'isStopWord': False}, 'meeting': {'occurance': 2, 'isStopWord': False}, 'graphical': {'occurance': 1, 'isStopWord': False}, 'easy': {'occurance': 2, 'isStopWord': False}, "That's": {'occurance': 2, 'isStopWord': True}, 'And': {'occurance': 6, 'isStopWord': True}, 'has': {'occurance': 1, 'isStopWord': True}, 'look': {'occurance': 1, 'isStopWord': True}, 'these': {'occurance': 1, 'isStopWord': True}, 'tastes': {'occurance': 1, 'isStopWord': False}, 'will': {'occurance': 3, 'isStopWord': True}, 'while': {'occurance': 1, 'isStopWord': True}, 'fun': {'occurance': 1, 'isStopWord': False}, 'PowerPoint': {'occurance': 2, 'isStopWord': False}, 'picturing': {'occurance': 1, 'isStopWord': False}, 'pack': {'occurance': 1, 'isStopWord': False}, 'almost': {'occurance': 1, 'isStopWord': True}, 'is': {'occurance': 14, 'isStopWord': True}, 'it': {'occurance': 23, 'isStopWord': True}, 'experts': {'occurance': 1, 'isStopWord': False}, 'Pretty': {'occurance': 1, 'isStopWord': False}, 'in': {'occurance': 8, 'isStopWord': True}, 'You': {'occurance': 1, 'isStopWord': True}, 'casings': {'occurance': 1, 'isStopWord': False}, 'if': {'occurance': 10, 'isStopWord': True}, 'different': {'occurance': 2, 'isStopWord': True}, 'shorter': {'occurance': 1, 'isStopWord': False}, 'vegetable': {'occurance': 1, 'isStopWord': False}, 'About': {'occurance': 1, 'isStopWord': True}, 'used': {'occurance': 3, 'isStopWord': True}, 'temporary': {'occurance': 1, 'isStopWord': False}, 'I': {'occurance': 27, 'isStopWord': True}, 'Maybe': {'occurance': 1, 'isStopWord': True}, 'running': {'occurance': 1, 'isStopWord': False}, 'fruit': {'occurance': 1, 'isStopWord': False}, 'purpose': {'occurance': 1, 'isStopWord': False}, 'recent': {'occurance': 1, 'isStopWord': True}, 'off': {'occurance': 2, 'isStopWord': True}, 'i': {'occurance': 1, 'isStopWord': True}, 'colour': {'occurance': 3, 'isStopWord': False}, 'fashionable': {'occurance': 1, 'isStopWord': False}, 'well': {'occurance': 2, 'isStopWord': True}, 'It': {'occurance': 2, 'isStopWord': True}, 'obviously': {'occurance': 1, 'isStopWord': True}, 'command': {'occurance': 1, 'isStopWord': False}, 'Have': {'occurance': 1, 'isStopWord': True}, 'y': {'occurance': 1, 'isStopWord': True}, 'the': {'occurance': 37, 'isStopWord': True}, 'audio': {'occurance': 1, 'isStopWord': False}, 'know': {'occurance': 8, 'isStopWord': True}, 'just': {'occurance': 7, 'isStopWord': True}, 'true': {'occurance': 1, 'isStopWord': False}, 'batteries': {'occurance': 1, 'isStopWord': False}, 'aspect': {'occurance': 3, 'isStopWord': False}, 'olds': {'occurance': 1, 'isStopWord': False}, 'previous': {'occurance': 1, 'isStopWord': False}, 'web': {'occurance': 1, 'isStopWord': False}, 'thinking': {'occurance': 1, 'isStopWord': False}, "we've": {'occurance': 4, 'isStopWord': True}, 'seems': {'occurance': 1, 'isStopWord': True}, 'simple': {'occurance': 1, 'isStopWord': False}, 'Although': {'occurance': 1, 'isStopWord': True}, 'relevance': {'occurance': 2, 'isStopWord': False}, 'look-and-feel': {'occurance': 2, 'isStopWord': False}, 'aspects': {'occurance': 1, 'isStopWord': False}, 'around': {'occurance': 1, 'isStopWord': True}, 'Oh': {'occurance': 6, 'isStopWord': True}, 'Mono': {'occurance': 1, 'isStopWord': False}, 'gonna': {'occurance': 2, 'isStopWord': False}, 'five': {'occurance': 4, 'isStopWord': True}, 'T': {'occurance': 1, 'isStopWord': True}, 'bit': {'occurance': 2, 'isStopWord': False}, 'Or': {'occurance': 3, 'isStopWord': True}, 'furniture': {'occurance': 1, 'isStopWord': False}, 'insert': {'occurance': 1, 'isStopWord': False}, 'like': {'occurance': 5, 'isStopWord': True}, 'success': {'occurance': 1, 'isStopWord': False}, 'specific': {'occurance': 1, 'isStopWord': False}, 't': {'occurance': 1, 'isStopWord': True}, 'Again': {'occurance': 1, 'isStopWord': True}, 'because': {'occurance': 2, 'isStopWord': True}, 'often': {'occurance': 1, 'isStopWord': True}, 'people': {'occurance': 2, 'isStopWord': False}, 'some': {'occurance': 3, 'isStopWord': True}, 'back': {'occurance': 2, 'isStopWord': True}, 'trendy': {'occurance': 1, 'isStopWord': False}, 'New': {'occurance': 1, 'isStopWord': True}, 'scale': {'occurance': 1, 'isStopWord': False}, 'for': {'occurance': 6, 'isStopWord': True}, 'affects': {'occurance': 1, 'isStopWord': False}, 'face-plates': {'occurance': 3, 'isStopWord': False}, '?': {'occurance': 12, 'isStopWord': False}, 'Something': {'occurance': 1, 'isStopWord': True}, 'be': {'occurance': 11, 'isStopWord': True}, 'power': {'occurance': 1, 'isStopWord': False}, 'marketing': {'occurance': 1, 'isStopWord': False}, 'Um': {'occurance': 15, 'isStopWord': False}, 'Command': {'occurance': 1, 'isStopWord': False}, 'Uh': {'occurance': 2, 'isStopWord': False}, 'Think': {'occurance': 1, 'isStopWord': True}, 'on': {'occurance': 10, 'isStopWord': True}, 'about': {'occurance': 4, 'isStopWord': True}, 'actual': {'occurance': 1, 'isStopWord': False}, 'working': {'occurance': 1, 'isStopWord': True}, 'getting': {'occurance': 1, 'isStopWord': True}, 'of': {'occurance': 22, 'isStopWord': True}, 'discussed': {'occurance': 1, 'isStopWord': False}, 'draf': {'occurance': 1, 'isStopWord': False}, 'wanna': {'occurance': 1, 'isStopWord': False}, 'or': {'occurance': 3, 'isStopWord': True}, 'functional': {'occurance': 1, 'isStopWord': False}, 'right': {'occurance': 2, 'isStopWord': True}, 'been': {'occurance': 3, 'isStopWord': True}, 'your': {'occurance': 4, 'isStopWord': True}, 'her': {'occurance': 1, 'isStopWord': True}, 'vegetables': {'occurance': 1, 'isStopWord': False}, 'there': {'occurance': 4, 'isStopWord': True}, "we're": {'occurance': 3, 'isStopWord': True}, 'lot': {'occurance': 3, 'isStopWord': False}, 'was': {'occurance': 9, 'isStopWord': True}, 'spongy': {'occurance': 1, 'isStopWord': False}, 'Well': {'occurance': 2, 'isStopWord': True}, 'but': {'occurance': 2, 'isStopWord': True}, 'idea': {'occurance': 2, 'isStopWord': False}, 'volume': {'occurance': 1, 'isStopWord': False}, 'granny': {'occurance': 1, 'isStopWord': False}, 'kitchenware': {'occurance': 1, 'isStopWord': False}, "I'll": {'occurance': 1, 'isStopWord': True}, "they're": {'occurance': 1, 'isStopWord': True}, 'up': {'occurance': 5, 'isStopWord': True}, 'us': {'occurance': 1, 'isStopWord': True}, 'um': {'occurance': 6, 'isStopWord': False}, 'strawberry': {'occurance': 1, 'isStopWord': False}, 'uh': {'occurance': 7, 'isStopWord': False}, 'called': {'occurance': 1, 'isStopWord': False}, 'sales': {'occurance': 1, 'isStopWord': False}, 'an': {'occurance': 4, 'isStopWord': True}, 'as': {'occurance': 6, 'isStopWord': True}, 'at': {'occurance': 2, 'isStopWord': True}, 'Real': {'occurance': 1, 'isStopWord': False}, 'ideally': {'occurance': 1, 'isStopWord': False}, 'when': {'occurance': 1, 'isStopWord': True}, 'rubber': {'occurance': 3, 'isStopWord': False}, 'other': {'occurance': 1, 'isStopWord': True}, 'you': {'occurance': 19, 'isStopWord': True}, 'Like': {'occurance': 3, 'isStopWord': True}, 'favourite': {'occurance': 1, 'isStopWord': False}, 'update': {'occurance': 1, 'isStopWord': False}, 'important': {'occurance': 5, 'isStopWord': True}, 'kinda': {'occurance': 1, 'isStopWord': False}, 'sides': {'occurance': 1, 'isStopWord': True}, 'remote': {'occurance': 7, 'isStopWord': False}, 'T_V_': {'occurance': 2, 'isStopWord': False}, 'age': {'occurance': 1, 'isStopWord': False}, 'Yeah': {'occurance': 13, 'isStopWord': False}, 'time': {'occurance': 1, 'isStopWord': False}, 'everywhere': {'occurance': 1, 'isStopWord': True}, 'once': {'occurance': 2, 'isStopWord': True}, 'having': {'occurance': 1, 'isStopWord': True}, 'basically': {'occurance': 1, 'isStopWord': False}}
Content source: boompieman/iim_project
Similar notebooks: