Use chinease natual language processing library NLPIR-ICTCLAS to do the work like Chinease word segmentation and part-of-speech tagging. Preprocessing. Use word2vec gensim to train the model and caculate the similarity.