类型错误：“Word2Vec”对象不可下标

Question

我正在尝试构建 Word2vec 模型，但是当我尝试重塑标记向量时，出现此错误。任何想法 ？

wordvec_arrays = np.zeros((len(tokenized_tweet), 100)) 
for i in range(len(tokenized_tweet)):
    wordvec_arrays[i,:] = word_vector(tokenized_tweet[i], 100)
wordvec_df = pd.DataFrame(wordvec_arrays) 
wordvec_df.shape

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-101-71156bf1c4a3> in <module>
      1 wordvec_arrays = np.zeros((len(tokenized_tweet), 100))
      2 for i in range(len(tokenized_tweet)):
----> 3     wordvec_arrays[i,:] = word_vector(tokenized_tweet[i], 100)
      4 wordvec_df = pd.DataFrame(wordvec_arrays)
      5 wordvec_df.shape

<ipython-input-100-e3a82e60af93> in word_vector(tokens, size)
      4     for word in tokens:
      5         try:
----> 6             vec += model_w2v[word].reshape((1, size))
      7             count += 1.
      8         except KeyError: # handling the case where the token is not in vocabulary

TypeError: 'Word2Vec' object is not subscriptable

Answer 1

从 Gensim 4.0 及更高版本开始，该Word2Vec模型不支持下标索引访问（该['...']') to individual words. (Previous versions would display a deprecation warning, 方法将在 4.0.0 中删除，使用 self.wv.getitem ( ) 代替`，用于此类用途。）

因此，当您想要访问特定单词时，请通过Word2Vec模型的.wv属性来实现，该属性仅包含单词向量。因此，您的（未显示）word_vector()函数应该将错误堆栈中突出显示的行更改为：

            vec += model_w2v.wv[word].reshape((1, size))