In recent years, social software such as matchmaking websites and dating platforms have gained popularity, reflecting the plight of young people in expanding their real-life social circle and finding their ideal partner. However, the current popular social platforms have obvious problems in matching quality and stability, as well as insufficient user privacy protection. To address this, we propose a marriage matching for bipartite graphs based on Condensed Local Differential Privacy. Specifically, we propose multiple data perturbation mechanisms and interactive user opinion collection aggregation methods to ensure privacy protection while improving data utility. In addition, we optimize the Gale-Shapley algorithm to improve the accuracy and robustness of matching. Experiments demonstrate that our scheme can converge to the matching results of real data on several satisfaction metrics, proving its usability and privacy.