This sentence similarity calculation is based on multi-feature and word vector, combining sentence surface information, dependent structure and semantics.
Using senteces from Sougo News library
Jieba Text Segmentation, Word2Vec, StanfordParser, NLTK
Automatic Text Summarization