Optimizing researcher mentorship matching: A particle swarm optimization-based recommendation model

Mentoring is an essential collaborative practice among academic researchers, fostering growth and expertise. It is widely believed that scientific knowledge, practices, and skills are transferred from one generation of scientists to the next through mentorship. The increasing significance of collaboration among academic researchers necessitates innovative, effective tools for optimal mentor-mentee matching, facilitating successful mentorship and knowledge transfer. Despite existing expert-finding recommender systems, matching mentors with mentees remains understudied. This research addresses this gap by developing a novel metaheuristic-based approach to optimize mentor-mentee pairing. Utilizing profile and publication datasets from Academic Family Tree, a Support Vector Machine (SVM) classifier is employed to categorize researchers as experts or young researchers. Term Frequency-Inverse Document Frequency (TF-IDF) extracts research area features, generating researcher vectors. These inputs are then optimized using Particle Swarm Optimization (PSO) algorithm to facilitate mentorship connections. The results demonstrate exceptional performance: the Support Vector Machine (SVM) classifier achieves 99% accuracy, while the optimized recommendation model based on PSO algorithm, which achieves 100% accuracy, outperforms three baseline models, collaborative filtering (CF), content-based filtering (CBF) and Hybrid CF-CBF models. This study's findings can inform research institutions seeking to enhance researcher-mentor connections, fostering collaborative excellence. Future research will explore expanded datasets and algorithmic refinements.