Evaluating an Intelligent Tutoring System for Personalized Math Teaching

Abstract

With the popularity of Intelligent Tutoring System (ITS), it is important to evaluate its teaching effect, but such kind of evaluation is rarely found in Chinese research community. We studied "Lexue 100", a personalized and web-based math-teaching ITS. In the study, we use quasi-experiment method to evaluate students' performances, so the experiment class and control class are set to make internal and external comparison. The examination score analysis shows that the system can significantly improve the math performance of the experiment class in regular school examinations. The collected students' answers are reliable and valid, and their statistical analysis demonstrates that this system's teaching quality is overall good including the interface, content, instant feedback and game reward mechanism, which can be very helpful for the students' math learning. The special word frequency analysis also proves that most students find this system to be positive and beneficial to their learning, but functional improvement is still needed.