求帮忙跑一下，实在找不出问题在哪，基于word2vec的推荐系统

Question

基于word2vec的推荐系统named item2vec。网上也能搜到，想实现一下。文章中准确率达0.44，我跑出来的低的令人发指，是在找不出问题在哪，有兴趣的希望能帮忙跑一下，谢谢！；python 2.7

from gensim.models import Word2Vec   
import logging  
import sys
reload(sys)
sys.setdefaultencoding('utf8')
from sklearn.model_selection import train_test_split
c = []

def load_sequence(from_path):
    with open(from_path) as fp:
        [c.append(line.strip().split(",")) for line in fp]
        
def main():
    logging.basicConfig(format='%(asctime)s : %(levelname)s : %(message)s', level=logging.INFO)  
    load_sequence('E:\\pythonwork\\1105\\to666.txt') # 加载语料  
    c_train,c_text = train_test_split(c,test_size=0.2)
    model = Word2Vec(c_train, size=20, window=3, min_count=1, workers=1, iter=3, sample=1e-4, negative=20)  # 训练skip-gram模型; 默认window=5  
    test_size = float(len(c_text))
    hit = 0.0
    for current_pattern in c_text:
        if len(current_pattern) < 2:
            test_size -= 1.0
            continue
        # Reduce the current pattern in the test set by removing the last item
        last_item = current_pattern.pop()

        # Keep those items in the reduced current pattern, which are also in the models vocabulary
        items = [it for it in current_pattern if it in model.wv.vocab]
        if len(items) <= 2:
            test_size -= 1.0
            continue

        # Predict the most similar items to items
        prediction = model.most_similar(positive=items,topn=20)

        # Check if the item that we have removed from the test, last_item, is among
        # the predicted ones.
        for predicted_item, score in prediction:
            if predicted_item == last_item:
                hit += 1.0
    print 'Accuracy like measure: {}'.format(hit / test_size)

if __name__ == "__main__":  
    main()

数据集：链接描述

参考样例：链接描述