1. Remove the first line into Content []
= Contents []
Contents = df.ix [:, 10] of 11 data #
2. extracted data by type
good = df.loc [df [ "evaluation type"] == "Good"] good_contents good.ix = [:, 10] good_contents.index = List (Range (good_contents.shape [0])) re-index #
3. Extract Keywords
textrank DEF (Contents, TOPK):
cons = []
for I in Range (len (Contents)):
Content = Contents [I]
Keywords = jieba.analyse.textrank (Content, TOPK = TOPK, allowPOS = ( 'n-', 'nz', 'v', 'vd', 'vn', 'l', 'a', 'd')) # TextRank keyword extraction, filtering speech
word_split = " ".join(keywords)
#print(word_split)
cons.append(word_split) #.encode("utf-8")
result = pd.DataFrame(cons)
return result
4. The extracted keyword output file
result.to_csv("keywords.csv", encoding = 'utf-8',index=False)
5. Problems
When the content que :: textrank algorithm will first inputted word, after the first stage will each comment text extracted from the second stage to extract all keywords keywords, between words have been separated by spaces . This is reasonable?
- "directly or based on word frequency, maximum output frequency of keywords
que: When do the second stage, the first set of keywords directly with spaces put together? This reasonable?
--"to be solved
- "adjectives, adverbs do not output
good = df.loc[df["评价类型"]=="好评"]
good_contents = good.ix[:,10]
good_contents.index=list(range(good_contents.shape[0]))