Get 50 chinese news from test data,find each of them the most similar 50 news in train data. enviroment win 10 anaconda 4.3.1 python 3 jieba 0.39 gensim 3.4.0 TODO calculate the sentence-meaning vector for each news to cut down the calculated quantity. cluster the news to implement TFIDF on each news to find out the key words. demo cmd python get_top20.py