Host: The Japanese Society for Artificial Intelligence
Name : The 38th Annual Conference of the Japanese Society for Artificial Intelligence
Number : 38
Location : [in Japanese]
Date : May 28, 2024 - May 31, 2024
When creating questions for mock tests such as high school entrance exams, it is desirable to appropriately adjust the difficulty level of the questions so that the student's level of understanding is reflected in the score. Traditionally, this adjustment of difficulty has been done by test creators by referring to past questions, but if the correct answer rate for new questions can be predicted with high accuracy, it would be helpful in adjusting the difficulty of tests. In this paper, we use a method that predicts the correct answer rate for new questions based on the question features extracted from past questions that the author has verified for social studies and science tests, and develop a method for predicting the percentage of correct answers to Japanese long text reading comprehension questions, and verify how accurately the percentage of correct answers can be predicted.
