conv_net_sentence.py 文件源码

def make_idx_data_cv(revs, word_idx_map, cv, max_l=51, k=100, filter_h=5):
    """
    Transforms sentences into a 2-d matrix.
    """
    train, test = [], []
    for rev in revs:
        sent = get_idx_from_sent(rev["text"], word_idx_map, max_l, k, filter_h)   
        sent.append(rev["y"])
        if rev["split"]==cv:            
            test.append(sent)        
        else:  
            train.append(sent)   
    train = np.array(train,dtype="int")
    train = np.random.permutation(train)[:10000]
    test = np.array(test,dtype="int")
    test = np.random.permutation(test)[:10000]
    return [train, test]