Presentation information


General Session

General Session » GS-10 AI application

[4F1-GS-10l] AI応用:製品の品質

Fri. Jun 11, 2021 9:00 AM - 10:40 AM Room F (GS room 1)

座長:植野 研(東芝)

9:00 AM - 9:20 AM

[4F1-GS-10l-01] Multilevel tree compression for adaptive testing using decision trees to reduce generation time

〇Naoki Akasaka1, Kazuma Fuchimoto1, Maomi Ueno1 (1. The University of Electro-Communications)

Keywords:adaptive testing, e-testing

Recently, Computer Adaptive Testing (CAT) based on decision tree has attracted a broad attention for educational assessment. However, CAT using decision tree has the problem that the amount of time and space complexities increase as the number of branches increases exponentially. To solve this problem, a method called Merged Tree-CAT has been proposed to reduce the number of branches by merging the nodes that have similar distribution of the estimated abilities. Although this method reduces the generation time of decision tree, it still requires computational time cost when the item bank is large or the number of items is large because it merges nodes which are only the same level of decision tree. In this study, we proposed a method which merges nodes at the same or higher levels of decision tree. As a result, we succeed to reduce the generation time of decision tree without deterioration in estimation accuracy.

Authentication for paper PDF access

A password is required to view paper PDFs. If you are a registered participant, please log on the site from Participant Log In.
You could view the PDF with entering the PDF viewing password bellow.