An IRT-Based Approach to Obtaining Item-Aware Learning Achievement

Due to the technical advances of computer networks, developing a learning assessment system with the diagnosis technology induced has become an important issue. As we know, data mining is a powerful tool in discovering hidden knowledge or rules such as Concept Effect Relationships from the testing results. However, since the learning achievement decided by the testing results may be affected by the profile of the test item, analyzing the learning achievement derived directly from the testing results without the preprocessing may cause the mining ineffective. Therefore, Item-Aware Learning Achievement, which considers the effect of the test item profile, is needed to solve this issue. In this paper, we first propose an IRT (Item Response Theory) Based Approach to obtain the Item-Aware Learning Achievement, and then apply the Item-Aware Learning Achievement in constructing the concept map to evaluate the performance of the IRT-Based Approach. The experimental results show that the Item-Aware Learning Achievement obtained by the proposed approach can improve the efficiency of the concept map construction.