Based on the voting framework of complex networks, an evaluation function of global optimal winners, which can act as a criterion for evaluating the decision of the Condorcet rule and the majority approval rule, is established. By introducing voters'point supporting region and setting the effectiveness decision condition for the optimal candidate, we construct a “compromise-support” index. With the different priority between compromise and support, we can solve the search solution for the global optimal candidate using a global optimization algorithm by the Delaunay triangle partition, and present the evaluation function to measure some voting rules. Numerical simulations show that, in most networks, the proposed evaluation function can measure more accurately the deviation degree to people's real willing for the Condorcet rule and the majority approval rule.