使用WordNet和NLTK在语料库中替换同义词 - python

Question

我想写一个简单的Python脚本，利用NLTK这个库来在文本文件中查找并替换同义词。

下面的代码让我遇到了错误：

Traceback (most recent call last):
  File "C:\Users\Nedim\Documents\sinon2.py", line 21, in <module>
    change(word)
  File "C:\Users\Nedim\Documents\sinon2.py", line 4, in change
    synonym = wn.synset(word + ".n.01").lemma_names
TypeError: can only concatenate list (not "str") to list

这是代码：

from nltk.corpus import wordnet as wn

def change(word):
    synonym = wn.synset(word + ".n.01").lemma_names

    if word in synonym:

            filename = open("C:/Users/tester/Desktop/test.txt").read()
            writeSynonym = filename.replace(str(word), str(synonym[0]))
            f = open("C:/Users/tester/Desktop/test.txt", 'w')
            f.write(writeSynonym)
            f.close()

f = open("C:/Users/tester/Desktop/test.txt")
lines = f.readlines()

for i in range(len(lines)):

    word = lines[i].split()
    change(word)

Answer 1

有两件事。首先，你可以把读取文件的部分改成：

for line in open("C:/Users/tester/Desktop/test.txt"):
    word = line.split()

其次，.split() 会返回一个字符串列表，而你的 change 函数似乎只处理一个单词。这就是导致错误的原因。你的 word 实际上是一个列表。

如果你想处理那一行中的每一个单词，可以把它改成：

for line in open("C:/Users/tester/Desktop/test.txt"):
    words = line.split()
    for word in words:
        change(word)

Answer 2

这样做效率不是特别高，而且这并不能替换一个单独的同义词，因为每个词可能有多个同义词。你可以从中选择一个。

from nltk.corpus import wordnet as wn
from nltk.corpus.reader.plaintext import PlaintextCorpusReader


corpus_root = 'C://Users//tester//Desktop//'
wordlists = PlaintextCorpusReader(corpus_root, '.*')


for word in wordlists.words('test.txt'):
    synonymList = set()
    wordNetSynset =  wn.synsets(word)
    for synSet in wordNetSynset:
        for synWords in synSet.lemma_names:
            synonymList.add(synWords)
    print synonymList