Timely graduation is an important indicator of the quality of higher education. Yet, many students struggle to complete their studies on time due to challenges in finding relevant research topics and suitable supervisors. This study developed a two-way supervisor recommendation system that considers the preferences and expertise of both students and supervisors. The main contribution of this research is the comparison of Block Nested Loop (BNL) k-skyband and MapReduce k-skyband algorithms. The recommendation model developed in this study uses course syllabi to obtain research topics and academic grades to determine students' interests in research topics. A total of 239 research topics were obtained from 37 courses. Optimal recommendations were achieved with a k value of 16. Implementing the MapReduce algorithm in this recommendation model demonstrated a computation speed 8 times faster than the BNL k-skyband approach, making it effective in handling large datasets. The proposed recommendation system received positive feedback from students, with scores of 3.5 for relevance, 3.7 for topic diversity, 3.4 for serendipity, and 3.5 for novelty. These findings suggest that the proposed recommendation system can support students in their research endeavors and improve the overall supervision process in academic settings, with potential for widespread implementation across various study programs. Thus, contributing to the overall improvement of higher education quality.