1. UKM Learning and Research Repository
2. Master Thesis / Tesis Master
3. Faculty of Information Science and Technology / Fakulti Teknologi dan Sains Maklumat

Title:	Exam questions classsification based on Bloom's taxonomy cognitive level using classifiers and rules
Authors:	Dhuha Abdulhadi Abduljabbar (P72232)
Supervisor:	Nazlia Omar, Assoc. Prof. Dr.
Keywords:	Exam questions Bloom's taxonomy Classsification Dissertations, Academic -- Malaysia
Issue Date:	8-Jun-2015
Description:	Assessment through written examination is a traditional and universal test method being practiced in most educational institutions today. This method requires that provided questions are coherent with the subject’s content learned by students in order to fulfill the learning objectives. However, the process of constructing the questions remains a challenge to the lecturer. The situation becomes increasingly challenging when lecturers attempt to generate high-quality and fair questions to assess different cognitive levels. Thus, Bloom’s taxonomy is a common reference used as an educational guide for the formulation of exam questions. Exam questions classification presents a particular challenge where it is the classification of short text questions. The questions can be considered as short text as it involves text with less than 200 characters, unlike text document. In addition, the presence of terms that contribute to the density of the text features is sparse. This study proposed a new method to classify exam questions automatically according to the cognitive levels of Bloom’s taxonomy. The proposed method is a combination strategy based on Voting algorithm that combines machine learning approaches and a rule-based approach. A number of experiments were conducted based on individual and combined classifiers methods. First, a rule-based approach is used to classify the question based on keywords, part of speech (POS) and shallow parsing information. Then, three machine learning classifiers, namely Support Vector Machine (SVM), Naive Bayes (NB), and K-Nearest Neighbour (KNN) are used to classify the question with or without feature selection methods, namely Chi-Square, Mutual Information and Odd Ratio. Finally, a combination algorithm is used to integrate the overall strength of the three classifiers (SVM, NB, and KNN) and the rule-based approach. The system used 195 of written examination questions for computer programming subjects as a training and test data set. The classification model achieves an average of 94.44% recall, 96.29% precision, and 94.58 % F1-measure through the combination strategy by applying Chi-Square feature selection. The experiments indicate that the combined approach yields the best classification method and outperforms the base classifiers.,Certification of Master's/Doctoral Thesis" is not available
Pages:	103
Publisher:	UKM, Bangi
Appears in Collections:	Faculty of Information Science and Technology / Fakulti Teknologi dan Sains Maklumat

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.