为什么建立embedding_matrix时要再一开始多加一行？

why : nb_words = len(tokenizer.word_index) + 1 ?????

answer: 

 

1. word index start from 1, so the index of 0 would be 0 all the time

embedding_matrix = np.zeros((nb_words, embedding_dim))

for word, i in word_index.items():
        try:
            embedding_vector = word_vectors[word]
            if embedding_vector is not None:
                embedding_matrix[i] = embedding_vector
        except KeyError:
            print("vector not found for word - %s" % word)

 

2. while mapping the word to vector, the padding of 0 would be mapping to np.array(0,embedding_dim)

    train_sequences_1 = tokenizer.texts_to_sequences(sentence1) # texts_to_sequences -------- map the index of words to words
    train_padded_data_1 = pad_sequences(train_sequences_1,maxlen = max_sequence_length) # padding to the max length of sequence
 

3. the sentence encoding would be complete