Similarity in formal argumentation has received some attention recently, since one can argue that, in some context, using similar arguments to reach a conclusion is not the same as using dissimilar ones. While existing work consider arguments built from propositional logic, in this work we adapt the notion of similarity measures to arguments built from Order-Sorted First Order Logic, an extension of First Order Logic which allows to represent complex information, considering the type of the data. We study and evaluate our approach with respect to an adaptation of axioms from the literature. This paves the way to new reasoning modes taking into account similarity between arguments in complex settings like ontologies.