An architecture for an adaptive assessment tool

The educational society faces the challenge of improving the quality of education, while increasing the percentage of students that pass a class, given limited teaching resources. In response to this demand, this work reports on the architecture and the implementation of a system for adaptive multi-user assessment of students. The software is a client/server platform developed in Java programming language, which is termed platform for adaptive and reliable evaluation of students (PARES). Hypertext markup language is used for the development of pages including text and figures. PARES includes three modules, which can be remotely connected to the system database on server through network. The Bayesian decision model is applied to adaptive testing, instead of the traditional item response theory. PARES can be used interactively by administrators, course instructors and students. When PARES is used for self-assessment, feedback is provided to the student´s answer. As a result, personalized learning assistance is provided to improve a student´s level of proficiency. Experimental results show that when students take self-assessment tests a week before scheduled formative tests, their scores are increased.