A collaborative recommender system for learning courses considering the relevance of a learner’s learning skills
| Autor: | Jaechoon Jo, Heuiseok Lim, Jiwon Han, Hyesung Ji |
|---|---|
| Rok vydání: | 2016 |
| Předmět: |
Multimedia
Computer Networks and Communications Computer science Learning environment 05 social sciences 050301 education Collaborative learning 02 engineering and technology Open learning Recommender system computer.software_genre Learning effect Active learning 0202 electrical engineering electronic engineering information engineering Collaborative filtering Profiling (information science) 020201 artificial intelligence & image processing 0503 education Curriculum computer Software |
| Zdroj: | Cluster Computing. 19:2273-2284 |
| ISSN: | 1573-7543 1386-7857 |
| DOI: | 10.1007/s10586-016-0670-x |
| Popis: | Recommender systems are needed in the educational environment, where different effects are observed depending on personal tendencies, and are currently applied to educational area with diverse methods. In particular, the recommender system is extremely useful in the non-formal learning environment in that it can provide differentiated learning courses according to learners' levels to improve the learning effects, reducing the difficulties, and supporting a trial-and-error approach in choosing courses. This paper proposes a collaborative recommender system, which can improve learning performance by recommending learning courses that are appropriate to users' learning level. The proposed recommender system, based on collaborative filtering, recommends learning courses through the developing a curriculum, student skill model and Delphi survey analysis in order to take the correlation between the learner's profiling and the learning skills into account. As a result of the analysis of the effects of the proposed recommender system, its mean value of satisfaction was higher by 0.6 than that of the collaborative filtering recommendation; its standard deviation value appeared to be lower by 0.17, signifying that only a few values did not approximate the mean value; furthermore, the kurtosis value was lower by 0.19, indicating a concentrated data distribution around the mean value. As a result, we were able to provide differentiated learning courses to users who are experiencing difficulties with a trial-and-error approach in choosing the learning courses and to obtain a result with improved satisfaction. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full text from SpringerLink