w

# -*- coding: utf-8 -*-
from nltk  import *

# TO FIX : No such file or directory
os.chdir(r‘E:\zpy‘)

f = open(‘reviews_text_lt_3.txt‘, ‘r‘)
f_r = f.read()
strList = f_r.split(‘ ‘)
fdist1 = FreqDist(strList)
#總的詞數
print fdist1
#表達式 keys()為我們提供了文本中所有不同類型的鏈表
vocabulary1 = fdist1.keys()
#通過切片看看這個鏈表的前 50 項
 

res0_50 =vocabulary1[:50]
print res0_50

C:\>python E:\zpy\wltp.py
<FreqDist with 16789 samples and 180043 outcomes>
[‘‘, ‘raining‘, ‘disappointing.It‘, ‘uncomfortable...‘, "lot‘s", ‘uv.\nSo,‘, ‘yellow‘, ‘Seller‘, ‘four‘, ‘vaporizers.I‘, ‘Does‘, ‘completely!!‘, ‘hanging‘, ‘Monday,‘, ‘asap!!This
 
‘, ‘Until‘, ‘instead.The‘, ‘malfunctioned.‘, ‘Lately‘, ‘looking‘, ‘LAST‘, ‘eligible‘, ‘electricity‘, ‘DISAPPOINTED‘, ‘oneWorks‘, ‘powdery‘, ‘unanswered‘, ‘also.‘, ‘refun
‘sooooo‘, ‘foul‘, ‘on\nafter‘, ‘fingers.‘, ‘advice:‘, ‘fingers,‘, ‘advice?‘, ‘each),‘, ‘month.I‘]

C:\>

SELECT
    amz_review_text
 
FROM
    amz_reviews_grab_us
WHERE
    amz_review_rating < 3
LIMIT 3000;

對於通過亞馬遜us美國站的買家而言，在數據庫前3000條的時間周期y-m-d內，在不考慮品類、價格、評分相對值等因素的情況下，

暫得出以下推測：
0-賣品屬性為yellow，其他條件相同情況下，可能不受歡迎，評分相對低；
1-周一可能會給買家糟糕的購買體驗，周一的促銷活動須結合其他因素，如人文風俗、新聞事件慎重；
註：dev的當前視角

nltp APP-分析買家評論的評分-高頻詞：二維關系