An effective recommendation framework for personal learning environments using a learner preference tree and a GA

Personalized recommendations are used to support the activities of learners in personal learning environments and this technology can deliver suitable learning resources to learners. This paper models the dynamic multipreferences of learners using the multidimensional attributes of resource and learner ratings by using data mining technology to alleviate sparsity and cold-start problems and increase the diversity of the recommendation list. The presented approach has two main modules: an explicit attribute-based recommender and an implicit attribute-based recommender. In the first module, a learner preference tree (LPT) is introduced to model the interests of learners based on the explicit multidimensional attributes of resources and historical ratings of accessed resources. Then, recommendations are generated by nearest neighborhood collaborative filtering (NNCF). In the second module, the weights of implicit or latent attributes of resources for learners are considered as chromosomes in a genetic algorithm (GA), and then this algorithm optimizes the weights according to historical ratings. Then, recommendations are generated by NNCF using the optimized weight vectors of implicit attributes. The experimental results show that the proposed method outperforms current algorithms on accuracy measures and can alleviate cold-start and sparsity problems and also generate a more diverse recommendation list.