DOI
https://doi.org/10.1109/ACCESS.2020.2977679

Summary
Augmented reality (AR) enables dynamically enriching the real environment with digital information and helps students to overcome the occurred learning difficulties. Thus, an image-based programming learning system with AR technology is introduced in this article. Deep learning recommendations and learning theory is applied to personalize learning path. The research is based on exploring the differences in computational thinking. That is, the research questions aim to identify differences in learning achievements and computational thinking if deep learning recommendations are used. Firstly, the authors describe the concept of AR, deep learning, and computational thinking in the education field. Secondly, the research methodology is introduced by describing the profile of participants, system, activity scenarios. Finally, the results of statistical analysis are presented to demonstrate that better learning outcomes were achieved by the group that used deep learning recommendations compared to the control group. The conclusion was supported by the qualitative analysis of interviews.

Relevance for Complex Systems Knowledge
The authors present an image-based programming learning system which incorporates augmented reality (AR) and deep learning recommendations. The deep learning recommendations were applied to suggest activities for the student and are based on the students’ behavior in the system such as time used to read material, time used to solve the assignment [mission], steps taken to solve the assignment. The authors state that the students under research were not from a department related to information technology therefore they had no or little experience in programming. However, they acknowledged that the presented AR learning system with the deep learning recommendations was helpful in getting programming knowledge, overcoming learning gaps and developing computational thinking skills, such as creativity, collaboration, critical thinking and others. Although this learning system was designed to learn a programming language, a similar approach [a learning system with deep learning recommendations] can be applied in other subjects.

Point of Strength
The authors state that deep-learning recommendations in an image-based programming learning system helps students to achieve higher results in computational thinking ability and learning a programming language compared to those who did not get deep-learning recommendations.