PERSONALIZED ITEM RECOMMENDATION BASED ON CLUSTERING ASSESSMENT KNOWLEDGE OF TEACHERS

Personalized item-recommendation enables computer-assisted assessment systems to make deliberate efforts to perform appropriate assessment strategies that fit the needs, purposes, preferences, and interests of teachers. To achieve this goal, researchers had investigated various techniques to help teachers compile their assessments or tests by improving the item recommendation mechanism of assessment systems. This research proposes a personalized item-recommendation mechanism that, by using assessment knowledge, clustering analysis, and collaborative filtering, automatically adapts itself to the assessment interests and preferences of different teachers. When teachers are compiling their assessments, this mechanism helps the teachers by recommending the best-fit test items available in the item banks and showing their allocation information. The mechanism is built into an e-learning system, named IKMAAS, built in one of our earlier projects. Firstly, the mechanism finds the clusters of similar teachers based on their assessment inclinations by using a cluster-mining method based on the K-means Clustering Algorithm. The inclinations, including assessment preferences and interests of the teachers on the topics of a course, are collected as assessment knowledge by the IKMAAS, Secondly, when a teacher is creating an assessment, the IKMAAS, based on the assessment knowledge of the cluster the teacher belonging to, make a personalized recommendation to the teacher both a proper proportions of the number of test items to the to-be-test course concepts and the best-fit test items available in the item banks for each course concept. To see the effectiveness of the personalized item-recommendation mechanism, an experiment was conducted, which invited eighteen teachers to use the IKMAAS system for editing their assessments for science courses of the third grade in elementary school in a period of three months and then asked the teachers to answer a questionnaire afterwards. Through their system usage logs and answers to the questionnaires, the results of the experiment shows that the item-recommendation mechanism built with assessment knowledge of clustered teachers can improve the item recommendation services effectively for an e-learning system. All the teachers are satisfied with the recommendations and agree that they actually reduce both pressure and load in compiling assessments. They are more than happy to share the knowledge of assessment in the way the system does.