Abstract:

Self-regulated learning (SRL) is a process that assists students in managing their actions, thoughts and emotions in order to benefit of successful learning experiences, by allowing them to strengthen their study skills, monitor their performance or their academic progress. SRL is strongly connected to motivation and consists, roughly, in three phases: forethought and planning (analyze the learning task, set goals towards completing the task), performance monitoring (enhance the progress towards the learning task, monitor the progress, monitor motivation for completing the task), and reflections on performance (evaluate performance on the learning task, manage emotional responses related to the learning experience). There are several strategies which can be applied to promote SRL, e.g.: setting goals, planning, attention control, help-seeking from their peers or tutor. There are several tools which support SRL, e.g. virtual tutors, educational recommender systems, but also computer adaptive testing applications. The current paper focuses on the design and development of a computer adaptive testing (CAT) system based on recent social and semantic web technologies which might be easily used in any university course to promote SRL and to help the students acquire the necessary knowledge level for passing the exams.

CAT is a type of assessment developed to increase the efficiency of estimating the examinee’s knowledge. In order to do this, the algorithm adjusts the difficulty of questions to the examinee level, based on his previous answers (tailored testing). With every following answered question, the computer estimates better the examinee’s knowledge. With this type of test, examinees are encouraged to demonstrate the high-est level of their performance. Our application proposes an adapted CAT algorithm. The test consists of multiple choice questions with four possible answers. The questions for the test are divided into five levels of difficulty. The score is calculated from the level of difficulty. The difficult questions earn more points than the easy ones. The examinees may answer the same percentage of questions correctly, but an examinee will gain a better score if they answered correctly to more difficult items. The test has a variable number of questions, depending on the examinee’s answers.

Besides presenting the innovative CAT system in the context of other SRL supporting tools, a case study related to the exploitation of the system in a Romanian technical university is also presented. More than 50 students, enrolled in a computer science program and in an electronics program, used the CAT tool to prepare themselves for the Programming Language course: they were allowed to use the tool for two weeks, as many times as they wanted. Learnograms (charts containing the learning curves of the students) were built and the performance progress was analyzed compared to the final grade obtained by the students at an exam with fixed content (all the students had the same subjects). Encouraging results related to the positive correlation between the use of CAT and the increase in students’ motivation and knowledge level were registered.

Keywords:

Self-regulated learning, computer adaptive testing, web-based learning, learnogram.