RESULTS OF IMPLEMENTING HYBRID CAT USING IRT AND NAÏVE BAYES
Keywords:
Item Response Model · Naive Bayes Model · CAT (Computer Adaptive Test) · 2 - Parameter Model · Recommendation SystemAbstract
With the rapid research happening in the field of education and testing many systems implementing the latest
techniques like Item response theory and machine learning have been conceived. In this paper we will look at
implementation of a system that aims at employing different fields of research to develop a comprehensive testing
platform. The system sets itself apart by attempting to be self sufficient such that it requires little or no human
intervention required. This is achieved through automatic question acquisition and classification through a community
run forum and a dynamic database that automatically transforms itself in accordance with the trends in the test takers
responses. It is based on Item response theory to get item characteristic classification for the question sets that allows for
an efficient test generator that can effectively test users across a larger section on latent scale. It also provides a
comprehensive result generation that informs the examinee about the various patterns in test thus allowing him to better
prepare.
