Design and Evaluation of an Adaptive KNN-Based Infant-Toddler Care Matching Platform

This study proposes an intelligent infant and childcare matching platform that integrates collaborative filtering with an Adaptive K-Nearest Neighbor (AKNN) algorithm to address information asymmetry, low efficiency in manual screening, and the difficulty of responding to diverse parental needs in traditional childcare matching services. The proposed system constructs a user perception matrix based on key attributes, including parents’ and caregivers’ available time, geographic location, professional qualifications, language ability, service fees, childcare preferences, and evaluation records. By combining cosine similarity with kernel density estimation, the AKNN model dynamically adjusts the value of K according to data distribution and neighborhood density, thereby generating more accurate matching scores and Top-N recommendations. The platform further improves recommendation quality by incorporating multiple dimensions of user preferences and caregiver characteristics. Simulation results indicate that the AKNN-based approach outperforms the conventional KNN method in key performance indicators, including accuracy, precision, recall, F1-score, and mean squared error. The findings suggest that the proposed platform can effectively reduce parents’ search costs, improve caregiver-resource allocation, alleviate the cold-start and data-sparsity problems, and enhance the overall efficiency and reliability of infant and childcare service matching.