Real time keyword extraction for e-learning system supporting quiz

In this paper, we propose a method to extract keywords from answers of a quiz. Teachers often do quizzes to grasp students´ misunderstandings. But, quizzes in a large class are hard for teachers, especially in a descriptive answer type. Teachers grasp misunderstandings from descriptive answers by following procedure: (1) skimming keywords, (2) grasping usage of keywords, (3) read answers if necessary. Supporting step (1) and (2), teachers´ burden would be reduced. We proposed the method extracting keywords by their frequency in answers and corpus. It is based on the idea: important words to answer a quiz are used frequently in its answers, and words also used frequently in corpus are common words, and not important. Since our method does not require complex analysis, it accept incomplete answers. Consequently, our method extracts keywords even during quizzes. With appropriate interface based on extracted keywords, teachers would grasp students´ misunderstandings quickly and can improve their classes just after quizzes.