Aligning users is to recognize accounts associated with the same individuals across different social networks. Currently, graph representation learning based and structural similarity based algorithms demonstrate their robustness and precision, respectively. Nevertheless, how to integrate different types of algorithms to enhance the alignment still needs to be addressed. To this end, we propose an integrated network alignment framework that augments the quality of alignment by heuristically determining which type of algorithm is more suitable for the alignment case by case. Specifically, given the alignment algorithm with varying strengths, a user-pair discriminator is designed with purposefully crafted features to predict the appropriate aligning algorithms for the users to be aligned and their corresponding candidate lists. Based on the discriminator, the proposed framework can be applied to a wide spectrum of alignment algorithms, and leverages the strengths of the corresponding algorithms to boost the performance of user alignment. Experiments conducted on real datasets have demonstrated that the proposed framework outperforms the counterpart single algorithm in terms of alignment precision.