首页 > 代码库 > nltp APP-分析买家评论的评分-高频词:二维关系

nltp APP-分析买家评论的评分-高频词:二维关系

w

 

 

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

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

f = open(reviews_text_lt_3.txt, r)
f_r = f.read()
strList = f_r.split( )
fdist1 = FreqDist(strList)
#总的词数
print fdist1
#表达式 keys()为我们提供了文本中所有不同类型的链表
vocabulary1 = fdist1.keys()
#通过切片看看这个链表的前 50 项
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., refunsooooo, 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;

 

对于通过亚马逊us美国站的买家而言,在数据库前3000条的时间周期y-m-d内,在不考虑品类、价格、评分相对值等因素的情况下,

暂得出以下推测:
0-卖品属性为yellow,其他条件相同情况下,可能不受欢迎,评分相对低;
1-周一可能会给买家糟糕的购买体验,周一的促销活动须结合其他因素,如人文风俗、新闻事件慎重;
注:dev的当前视角

 

nltp APP-分析买家评论的评分-高频词:二维关系