Identifying Optimal Baseline Variant of Unsupervised Term Weighting in Question Classification Based on Bloom Taxonomy

• Published in: Mendel (Brno (Czech Republic)) Vol. 28; no. 1; pp. 8 - 22
• Main Authors: Sangodiah, Anbuselvan, Jee San, Tham, Tien Fui, Yong, Ean Heng, Lim, Ayyasamy, Ramesh Kumar, A Jalil, Norazira
• Format: Journal Article
• Language: English
• Published: Brno University of Technology 30.06.2022
• ISSN: 1803-3814; 2571-3701
• DOI: 10.13164/mendel.2022.1.008

Examination is one of the common ways to evaluate the students’ cognitive levels in higher education institutions. Exam questions are labeled manually by educators in accordance with Bloom’s taxonomy cognitive domain. To ease the burden of the educators, several past research works have proposed the automated question classification based on Bloom’s taxonomy using the machine learning technique. Feature selection, feature extraction and term weighting are common ways to improve the accuracy of question classification. Commonly used term weighting method in the past work is unsupervised namely TF and TF-IDF. There are several variants of TF and TFIDF and the most optimal variant has yet to be identified in the context of question classification based on BT. Therefore, this paper aims to study the TF, TF-IDF and normalized TF-IDF variants and identify the optimal variant that can enhance the exam question classification accuracy. To investigate the variants two different classifiers were used, which are Support Vector Machine (SVM) and Naïve Bayes. The average accuracies achieved by TF-IDF and normalized TF-IDF variants using SVM classifier were 64.3% and 72.4% respectively, while using Naïve Bayes classifier the average accuracies for TF-IDF and normalized TF-IDF were 61.9% and 63.0% respectively. Generally, the normalized TF-IDF variants outperformed TF and TF-IDF variants in accuracy and F1-measure respectively. Further statistical analysis using t-test and Wilcoxon Signed also shows that the differences in accuracy between normalized TF-IDF and TF, TF-IDF are significant. The findings from this study show that the Normalized TF-IDF3 variant recorded the highest accuracy of 74.0% among normalized TF-IDF variants. Also, the differences in accuracy between Normalized TF-IDF3 and other normalized variants are generally significant, thus the optimal variant is Normalized TF-IDF3. Therefore, the normalized TF-IDF3 variant is important for benchmarking purposes, which can be used to compare with other term weighting techniques in future work.