Argumentation is a reasoning approach based on the justification of claims by arguments. It has been used for solving different Artificial Intelligence problems, including decision making, reasoning with defeasible information and classification. An argumentation framework is a directed graph, in which the nodes represent arguments, and the edges represent attacks between pairs of arguments. Gradual semantics are methods of evaluating arguments in graphs, that assign to each argument a numerical value, representing its strength. In this talk, I will present a general approach for defining gradual semantics for weighted graphs (where both arguments and attacks are weighted). I will show that several existing gradual semantics can be defined using the approach, and that it can be used to construct several novel semantics. The approach will be also analyzed against the set of properties from the literature.