A Kanji learning system based on automatic question sentence generation

Generally, people learn Kanji for The Japan Kanji Aptitude Test (commonly known as Kanken) by using workbooks. However, learners encounter the problems: the financial burden of workbook purchases and the insufficient volume of question collections. In order to solve these problems, we develop a Kanji learning system that can automatically generate question sentences of simulated Kanken tests based on Web search. After a user specifies the test grade and the number of questions, the system generates the specified number of question sentences in which Kanji of the specified grade are included. After the user answers the automatically generated questions, the system scores the answers and presents the learning results to the user. We conduct user studies on comparing three means of Kanji learning and the results show our Kanji learning system can help solve learners´ above-mentioned problems.