Abstract :
Recommender systems could be used to help learners or teachers find useful network teaching resources effectively in technology enhanced learning (TEL), but the quality of recommendations is always limited by cold start, data sparsity, lack of learning or teaching contextual aware, and so on. Considering the features of network teaching resources, a hybrid recommendation approach is presented in this paper. The presented approach takes user context, association rules between resources, association rules between resources and the structure of lessons into consideration, and is mainly composed of five modules. These five modules are: (1) Course model, which is used to express the structure of lessons; (2) Association rules between resources, which are discovered in resources association rule mining module; (3) Association rules between resources and lessons, which are discovered in lessons and resources association rule mining module; (4) User dynamic profile, namely, user context which are found in reasoning user dynamic profile module; (5) Hybrid recommendation, which generates recommended lists in hybrid recommendation module. Finally, experiments have been done on a real dataset from “HHT Education Cloud”, an enterprise education resources sharing platform. The results have shown that our hybrid method can outperform the general recommendation method. The Recall has an improvement ranging from 0.131 to 0.213, and the Precision has an improvement ranging from 0 to 0.152, when the number of recommendations changes from 1 to 40.
Keywords :
computer aided instruction; data mining; decision trees; educational courses; recommender systems; teaching; HHT Education Cloud; TEL; course model; enterprise education resource sharing platform; hybrid recommendation approach; knowledge-tree; lesson association rule mining module; lesson structure; network teaching resources; precision improvement; real dataset; recall improvement; recommended list generation; recommender systems; resource association rule mining module; technology enhanced learning; user context; user dynamic profile module reasoning; Association rules; Collaboration; Context; Dynamic scheduling; Education; Recommender systems; Association Rule Mining; Dynamic Profile; Knowledge-Tree; Recommender Systems;