[Objective] This paper proposes a new model to compute patent similarity, which fully
leverages the characteristics of patent texts and their structural and context …