The Intelligent Tutoring System AI-VT with Case-Based Reasoning and Real Time Recommender Models

This paper presents a recommendation model coupled on an existing CBR system model through a new modular architecture designed to integrate multiple services in a learning system called AI-VT (Artificial Intelligence Training System). The recommendation model provides a semi-automatic review of the CBR, two variants of the recommendation model have been implemented: deterministic and stochastic. The model has been tested with 1000 simulated learners, and compared with an original CBR system and BKT (Bayesian Knowledge Tracing) recommender system. The results show that the proposed model identifies learners’ weaknesses correctly and revises the content of the ITS (Intelligent Tutoring System) better than the original ITS with CBR. Compared to BKT, the results at each level of complexity are variable, but overall the proposed stochastic model obtains better results.