twitter_pos.py 文件源码

def embeddings_to_dict(filename):
    '''
    :param filename: the file name of the word embeddings | file is assumed
    to follow this format: "word[tab]dimension 1[space]dimension 2[space]...[space]dimension 50"
    :return: a dictionary with keys that are words and values that are the embedding of a word
    '''
    with io.open(filename, 'r', encoding='utf-8') as f:
        word_vecs = {}
        for line in f:
            line = line.strip('\n').split()
            word_vecs[line[0]] = np.array([float(s) for s in line[1:]])

    return word_vecs