Abstract: In most fields of Natural Language Processing (NLP), sentence similarity computing plays an important role. At the same time, the syntactic similarity computing is the basis of sentence similarity computing. This study introduces a dependency syntactic tree similarity computation method based on multi-features. The method studies many features of dependency syntactic tree including word and words POS of each node and the dependency type between them. Then the similarity algorithm is proposed after comprehensively analyzing all the features. The experimental result is satisfied as this method describes dependency syntactic tree more comprehensively and accurately.