Learning path recommender systems are emerging. Given the popularity of ontology/knowledge-based systems in adaptive learning, this work reviews learning path in ontology-based recommender systems. The review covers recommendation trends, ontology use, recommendation process, recommendation technique, contributing factors, and recommender evaluations. A total of 12,972 articles published between 2010 and 2020 were identified in the initial search across five major databases, and 9 of them are considered in this work. Currently, student model, learning objects, learning activities, and external environment are contributing factors for recommending learning object sequence. We also found that the current trend for LP recommendations process is semi-dynamic and dynamic. Semi-dynamic learning path are started by a pre-set path, while dynamic learning path is flexible from the first step and intended for personal use. The recommendation process itself has four phases: predelivery of the first learning object, current learning object delivery, learning object postdelivery, and predelivery of the next learning object. The current recommendation technique collaborates ontology and several techniques, such as Bayesian networks, data mining, and other artificial intelligence technique. To evaluate performance, learning path recommender systems use real students, control groups in parallel or sequential experiments, and student satisfaction surveys. Ontology could work with knowledge representation instruments, educational psychology, and evolutionary computation to create a future dynamic learning path in adaptive learning environment.