Page 86

www.rsisinternational.org

International Conference on Islamic Contemporary Issues and Management 2025

International Journal of Research and Innovation in Social Science (IJRISS) | ISSN: 2454-6186

DOI: 10.47772/I0JRISS | ICICM 2025 - Conference Proceedings | Volume IX Issue XXIX November 2025

An Intelligent Question Bank System for Automated Difficulty

Classification Based on Bloom’s Taxonomy

Md Razaul Karim

, Fatimah Noni Muhamad

, Mohd Zaki Shahabuddin

, Azhan Taqiyaddin Arizan

Noorkartina Mohamad

, Suharne Ismail

, Siti Khalilah Basarud-din

, Nurul Khofifah Abdullah

Nabilah Wafa’ Mohd Najib

, Naimah Abu Kassim

1,2,3,5,6

Faculty of Business and Management Science, Universiti Islam Antarabangsa Tuanku Syed

Sirajuddin, Malaysia

Faculty of Islamic Studies, Universiti Islam Antarabangsa Tuanku Syed Sirajuddin, Malaysia

7,8,9,10

Faculty of Muamalah and Islamic Finance, Universiti Islam Antarabangsa Tuanku Syed

Sirajuddin, Malaysia

DOI: https://dx.doi.org/10.47772/IJRISS.2025.92900016

Received: 17 November 2025; Accepted: 25 November 2025; Published: 17 December 2025

ABSTRACT

Creating and managing assessments is a challenging task for educators, especially when attempting to categorize

questions based on varying levels of difficulty. Traditional methods of question categorization are often done

manually, which takes a lot of time (time-consuming), and may lead to inconsistencies. These issues become

even more difficult when dealing with large question banks and inefficient administrative processes. To address

this, our research introduces the design and implementation of an Intelligent Question Bank System that

automates the classification of exam questions into difficulty levels : Easy, Medium, and Hard by using Bloom's

Taxonomy as the guiding framework. Bloom’s Taxonomy provides a hierarchical structure to categorize

cognitive skills, ranging from basic recall of facts to higher-order thinking skills like analysis and creation. The

system uses a Decision Tree algorithm, a type of Classification in Machine Learning, to classify questions based

on their complexity. This approach ensures accurate and consistent categorization by analyzing question

patterns, context, and semantics. The system is designed to handle large datasets effectively, making it a suitable

solution for educators managing extensive question banks. By combining Bloom’s Taxonomy with Machine

Learning techniques, the system simplifies the assessment process and improves its quality. It saves educators

time, helps them design better exams, and enhances the overall learning experience for students. This system

aims to transform the way questions are developed and managed, making education more efficient and effective.

Keywords: Intelligent Question Bank, Bloom Taxonomy, Classification, Machine Learning, Decision Tree

INTRODUCTION

Designing examination questions that accurately assess different levels of student understanding is a persistent

challenge for educators. Traditional methods of categorizing questions into easy, medium, and hard levels are

done manually, which often leads to inconsistent results and increased workload. These limitations can result in

unbalanced assessments and inefficiencies in managing large question banks. To address this, the Intelligent

Question Bank System is introduced as an AI-based solution that automates question classification using

Bloom’s Taxonomy, a trusted framework for organizing cognitive skills.

Bloom’s Taxonomy provides a structured approach to evaluating cognitive complexity, but applying it manually

across large datasets can be subjective and unreliable. Teacher judgments often vary, resulting in inconsistencies

in grading and misalignment with learning outcomes. To overcome these issues, the project integrates Machine

Learning—specifically a Decision Tree algorithm—to automatically classify questions into difficulty levels

based on semantic and contextual features aligned with Bloom’s hierarchy. This improves accuracy, reliability,

and fairness in assessment design.

Page 87

www.rsisinternational.org

International Conference on Islamic Contemporary Issues and Management 2025

International Journal of Research and Innovation in Social Science (IJRISS) | ISSN: 2454-6186

DOI: 10.47772/I0JRISS | ICICM 2025 - Conference Proceedings | Volume IX Issue XXIX November 2025

A key strength of the Intelligent Question Bank System is its ability to process extensive datasets quickly and

extensive datasets quickly and efficiently, something manual categorization cannot achieve. Decision Trees offer

an interpretable and effective method for automation, able to handle various types of data while providing

transparent reasoning for each classification. This reduces errors, saves time, and offers a scalable solution

suitable for institutions with large and diverse question repositories.

In addition to classification, the system includes reporting and analytics features that help educators understand

question distribution, performance, and effectiveness. These insights support data-driven decision-making,

ensuring assessments remain aligned with learning objectives. Overall, the Intelligent Question Bank System

advances educational assessment by combining artificial intelligence with established pedagogical frameworks,

offering a consistent, scalable, and evidence-based tool that enhances exam quality and supports improved

learning outcomes.

Problem Statement

Educators face significant challenges in designing and categorizing exam questions that accurately measure

student understanding across different difficulty levels. The manual classification of questions into Easy,

Medium, and Hard categories is time-consuming, subjective, and often inconsistent, leading to imbalanced

assessments. Managing large question banks across multiple subjects further increases this complexity and

administrative workload. Moreover, the lack of analytical tools makes it difficult for educators to evaluate

question effectiveness and track performance, hindering the development of data-driven, high-quality

assessments that enhance student learning outcomes.

Project Objectives

The objectives of this project are to develop an Intelligent Question Bank System that efficiently manages large

datasets and provides educators with a user-friendly interface to add, organize, and manage questions across

various subjects. The system aims to automate question classification based on Bloom’s Taxonomy, using a

Decision Tree Machine Learning algorithm to categorize exam questions into easy, medium, and hard levels

according to their complexity, context, and semantics, ensuring consistent and accurate results. Additionally, the

project seeks to enhance the assessment process through integrated reporting and analytics tools that offer

valuable insights into question usage, distribution, and effectiveness, enabling educators to make data-driven

improvements in assessment design and ultimately enhance the quality of education and student learning

outcomes.

Project Scope

The Intelligent Question Bank System focuses on creating an efficient and user-friendly platform that supports

multiple subjects and enables educators to easily create, organize, and manage questions. The system allows the

inclusion of additional materials such as images and supporting content to enhance question quality. Using AI-

powered algorithms based on Bloom’s Taxonomy, it will automatically classify questions into three difficulty

levels—Easy, Medium, and Hard—by analyzing their complexity and semantic patterns. Distinct user roles will

be defined for educators to manage and organize the question bank effectively. Additionally, integrated reporting

and analytics tools will provide insights into question usage and performance, helping educators improve the

quality and fairness of assessments..

LITERATURE REVIEW

An Intelligent Question Bank System (IQBS) based on comparing keywords is a software that allows users to

user (lecturer) to make a question based on Bloom Taxonomy. In addition to determine for the search accuracy,

these applications may also imply Natural Language Processing (NLP) which is a subfield of artificial

intelligence (AI) and filtering features that filter the relevant result searching.

The research paper by Chang (2021) proposes an automated system for selecting exam questions using a

binomial distribution model. The system aims to improve the efficiency and accuracy of creating test papers by

ensuring a balanced distribution of question difficulty and types. It ensures that each question has an equal chance

Page 88

www.rsisinternational.org

International Conference on Islamic Contemporary Issues and Management 2025

International Journal of Research and Innovation in Social Science (IJRISS) | ISSN: 2454-6186

DOI: 10.47772/I0JRISS | ICICM 2025 - Conference Proceedings | Volume IX Issue XXIX November 2025

of being selected, with criteria for difficulty level, question type, and knowledge points. The paper also discusses

the use of Expert-Based algorithms and Genetic algorithms for better question selection and large-scale test

paper creation. It highlights the importance of standardized test question banks to reduce teacher workload and

improve exam quality.

A research paper by Rachman, Alfian, and Yuhana (2023) focuses on using Deep Learning to classify questions

in the Indonesian National Assessment. It addresses the challenge of small datasets by applying data

augmentation and using Convolutional Neural Networks (CNN) to classify illustrated images into two

categories: information and literature. The study found that ResNet-50 gave the best performance, and transfer

learning helped solve the dataset issue. This work shows how Deep Learning can improve automated

classification in educational assessments, supporting Intelligent Question Bank Systems.

Yu, B. (2023) presents an intelligent system for creating exam papers using a Random Extraction algorithm.

This algorithm is selected for its speed and simplicity in meeting constraints like question type, quantity,

difficulty, and knowledge coverage. The study highlights the shift from traditional exams to AI-aided methods,

improving efficiency and fairness in assessments. It also compares the random extraction approach with other

algorithms and explores its application in teaching Chinese as a foreign language to enhance test quality.

The article by Zhang, J., and Li, Z. (2023) describes a question bank system using Machine Learning to support

intelligent paper grouping, online mock exams, and automated grading. Built with JSP, Java, and MySQL, it

includes roles for administrators, teachers, and students, ensuring fair and accurate marking by integrating

natural language processing and text similarity detection libraries. The system minimizes grading errors,

enhances personalized learning, and streamlines exams, reducing educators' workload while improving

assessment accuracy.

The research paper by Liu, X., and Yang, C. (2022) introduces an intelligent system for creating English test

papers using the Radial Basis Function (RBF) algorithm. The RBF neural network, known for its simple structure

and fast learning, helps select and organize questions based on knowledge points, difficulty levels, and total

scores. This system aims to reduce teachers’ workload and improve fairness in exams by ensuring a standardized

and efficient test composition process. It also addresses challenges in China’s paper-making systems, such as

inefficient grouping and the need for a large question bank, using advanced clustering and optimization

techniques. This innovative approach enhances the quality and accuracy of assessments in education.

Banujan et al. (2023) present an automated system for classifying over 16,000 exam questions from Sri Lankan

universities into the six levels of revised Bloom's Taxonomy (remember, understand, apply, analyze, evaluate,

create) using Deep Learning models like LSTM combined with BERT and GloVe embeddings, alongside ANN

with TF-IDF, achieving superior accuracy through semantic pattern recognition and NLP preprocessing such as

tokenization and stop-word removal. This approach addresses manual classification's subjectivity and scalability

issues in large question banks, outperforming traditional ML like SVM or DT by capturing contextual

relationships in text that directly complementing this project, which similarly analyzes keywords, complexity,

and cognitive demands for Easy/Medium/Hard categorization aligned with Bloom's framework, thus enhancing

consistency and efficiency in assessment design for educators managing extensive repositories.

While, Romadhony et al. (2022) present a Machine Learning framework for classifying primary and high school

exam questions into Bloom's Taxonomy's six cognitive levels (remembering, understanding, applying,

analyzing, evaluating, creating), using feature extraction techniques like TF-IDF variants and word embeddings

to detect semantic complexity and reduce teacher workload in manual categorization. This also directly

complements this project, which automates difficulty classification (Easy, Medium, Hard) aligned with Bloom's

hierarchy for large-scale question banks, by demonstrating ML's efficacy in handling diverse educational

datasets and ensuring consistent, scalable assessments across subjects addressing similar challenges of

inefficiency and subjectivity in traditional methods.

METHODOLOGY

The Intelligent Question Bank System (IQBS) uses a well-structured database to manage questions, categorized

by subject, topic, and difficulty. The system supports roles for Teacher (create, edit/manage, and categorize

Page 89

www.rsisinternational.org

International Conference on Islamic Contemporary Issues and Management 2025

International Journal of Research and Innovation in Social Science (IJRISS) | ISSN: 2454-6186

DOI: 10.47772/I0JRISS | ICICM 2025 - Conference Proceedings | Volume IX Issue XXIX November 2025

questions) and Admin (who manage the system). The system leverages AI, specifically a Decision Tree

algorithm, to classify questions by difficulty (Easy, Medium, Hard) using Bloom’s Taxonomy, analyzing factors

like wording, knowledge domain, and required cognitive skills. Teachers can build questions, track student

performance, and access analytics for better learning outcomes. It ensures security and supports future expansion

for additional subjects.

Flow Chart

According to the Figure 1.0, The Intelligent Question Bank System (IQBS) begins with a user-friendly interface

where teachers can add, organize, and manage questions. These questions are stored in a database, categorized

by subject and topic. Teachers input new questions, and the system uses a Decision Tree algorithm to classify

them into difficulty levels (Easy, Medium, Hard) based on Bloom's Taxonomy. The algorithm analyzes the

question's wording or keywords, cognitive skills required, and complexity to assign accurate difficulty levels.

Once classified, the questions are saved back to the database for use in generating balanced questions tailored to

student needs.

The Decision Tree algorithm mimics human decision-making by breaking down questions into attributes, such

as keywords or the knowledge domain, and comparing them against predefined criteria from Bloom’s

Taxonomy. For instance, a question that requires simple recall is classified as "Easy," while one involving critical

analysis is "Hard." This automated classification ensures consistency and accuracy, far surpassing traditional

methods. The system also integrates reporting and analytics tools, enabling teachers to track question

effectiveness, analyze student performance trends, and refine their questions. This data-driven approach supports

educators in enhancing teaching strategies and improving learning outcomes.

Figure 1.0 : The design for the work flow of IQBS.

Page 90

www.rsisinternational.org

International Conference on Islamic Contemporary Issues and Management 2025

International Journal of Research and Innovation in Social Science (IJRISS) | ISSN: 2454-6186

DOI: 10.47772/I0JRISS | ICICM 2025 - Conference Proceedings | Volume IX Issue XXIX November 2025

Context Diagram

The context diagram in Figure 2.0 shows how Admins and Teachers interact with the system to create/manage

questions, and generate reports. The system uses a Decision Tree algorithm to classify questions based on

difficulty levels (Easy, Medium, Hard) by analyzing keywords, complexity, and cognitive skills required. When

a teacher creates a question, the algorithm processes the question by identifying keywords such as "define,"

"analyze," or "evaluate," which correspond to different cognitive levels in Bloom’s Taxonomy.

The Decision Tree uses these keywords and other attributes to categorize the question, helping organize the

question bank for creating balanced questions. Admins manage the system, user accounts, and question bank,

while Teachers focus on creating, editing, and categorizing questions, and using analytics to improve questions.

Figure 2.0 : The Context Diagram of IQBS for the interaction between Admin and Teacher within the system.

Use Case Diagram

As depicted in Figure 3.0, the use case diagram for the Intelligent Question Bank System (IQBS) delineates the

interactions between the system and its users: Admin and Teacher. Each user engages with the system through

various functionalities tailored to their roles.

Figure 3.0 : The Use Case of IQBS for their functionalities.

The Admin holds full control over the system and interacts with several key functions. They begin by logging

in to authenticate their identity and gain access. Once inside, they manage the overall system settings, ensuring

smooth operation and proper configuration. After completing their tasks, the Admin logs out to maintain system

security.

Page 91

www.rsisinternational.org

International Conference on Islamic Contemporary Issues and Management 2025

International Journal of Research and Innovation in Social Science (IJRISS) | ISSN: 2454-6186

DOI: 10.47772/I0JRISS | ICICM 2025 - Conference Proceedings | Volume IX Issue XXIX November 2025

The Teacher focuses on managing questions and assessments within the system. They start by logging in to

access their assigned features. Teachers can create, edit, or delete questions to maintain the question bank. They

also view question analytics to understand performance patterns and generate detailed reports based on question

outcomes. When finished, they log out to ensure secure access and protect system data.

Entity Relationship Diagram

The Entity-Relationship Diagram (ERD) in Figure 4.0 represents the key entities and their relationships within

the system, focusing on Admin and Teacher roles. The Admin manages the system, including user accounts,

questions, and reports, ensuring the smooth operation of the question bank. The Teacher creates questions, and

the system uses an AI-based Decision Tree algorithm to analyze the difficulty level of each question by

evaluating its complexity and keywords. This classification is linked to Bloom's Taxonomy, assigning difficulty

levels (Easy, Medium, Hard). The Teacher then uses the classified questions to create questions and generate

reports, which provide insights into question usage and performance. The ERD connects these processes by

establishing relationships between entities such as User, Question, Difficulty Level, Assessment, and Report.

Figure 4.0 : The Entity-Relationship Diagram (ERD) of IQBS for the key entities and relationships within the

system.

Here's a detailed explanation of the components :

Entities and Attributes:

 Teacher

Attributes: TeacherID, Name, Email, Password

 Admin

Attributes: AdminID, Name, Email, Password

 Question

Attributes: QuestionID, Content, Subject, Difficulty

 Report

Attributes: ReportID, QuestionID, Score, Feedback

Page 92

www.rsisinternational.org

International Conference on Islamic Contemporary Issues and Management 2025

International Journal of Research and Innovation in Social Science (IJRISS) | ISSN: 2454-6186

DOI: 10.47772/I0JRISS | ICICM 2025 - Conference Proceedings | Volume IX Issue XXIX November 2025

Relationships:

 Manage (between Admin and Teacher)

Admins manage teachers, which involves adding, updating, or removing teacher records.

 Create/Analyze (between Teacher and Question)

Teachers create and analyze questions. They can add, edit, or evaluate the difficulty of questions.

 Generate (multiple occurrences)

Teachers generate questions by selecting questions and creating assessment documents.

 Questions generate reports for the difficulty levels.

Admins and teachers generate various types of reports to analyze performance and question usage.

RESULT

The results show how the question analysis feature examines question structure, content, and expected student

responses by identifying themes, assessing complexity, and predicting areas where students may struggle across

hard, medium, and easy categories. It also explains how these insights connect with the system’s automatic

difficulty classification to give a more complete understanding of each question.

The Frontpage of the Intelligent Question Bank System (IQBS)

Figure 5.0 showcases the homepage of the Intelligent Question Bank System (IQBS), designed for user-friendly

navigation and highlighting its core functionalities. The page features a prominent welcome message and three

interactive cards: "Create Questions," "View Questions," and "Analysis," each with a brief description and a

call-to-action button. Below, an "About Us" section explains the system's purpose, emphasizing personalized

learning, efficient content organization, and alignment with Bloom's Taxonomy to enhance teaching strategies

and support higher-order thinking. A left sidebar provides navigation links to the different sections of the IQBS,

including a "Sign Out" option for secure user sessions. The overall design aims to clearly communicate the

system's capabilities and facilitate easy access to its key features.

Figure 5.0 : The homepage of the Intelligent Question Bank System (IQBS), designed for user-friendly

navigation and highlighting its core functionalities.

Page 93

www.rsisinternational.org

International Conference on Islamic Contemporary Issues and Management 2025

International Journal of Research and Innovation in Social Science (IJRISS) | ISSN: 2454-6186

DOI: 10.47772/I0JRISS | ICICM 2025 - Conference Proceedings | Volume IX Issue XXIX November 2025

This project delves into the challenges faced by educators in creating and managing assessments, particularly

the time-consuming and often inconsistent process of classifying questions based on difficulty.

Classification base on Bloom taxonomy

Figure 6.0 displays the classification results of a set of questions based on Bloom's Taxonomy. Each

question is assigned a Bloom level (C1-C6) and a corresponding keyword that indicates the cognitive level

required to answer it. For example, questions involving "solve" or "solve the following" are classified under C3

(Applying), while questions asking to "Define" something fall under C1 (Remembering). Similarly, "Explain"

leads to "Understanding" at C2, and "Design" leads to C6 (Creating).

Figure 6.0 : Classification result of a set of questions based on Bloom's Taxonomy.

This table provides a structured overview of the cognitive demands of different questions, aiding assessment

design and curriculum planning.

CONCLUSION

The Intelligent Question Bank System (IQBS) serves as an advanced tool for managing and categorizing exam

questions efficiently by subject, topic, and difficulty level. Utilizing a Decision Tree algorithm aligned with

Bloom’s Taxonomy, it automatically classifies questions into Easy, Medium, and Hard categories based on

cognitive complexity. The system enables teachers to create, organize, and tailor questions to student needs while

offering integrated reporting and analytics tools to track performance and enhance learning outcomes.

Administrators oversee system security and scalability, ensuring support for future subject expansions. As

development continues, the IQBS aims to include JSU/JSI reporting features that will allow educators to analyze

syllabus effectiveness and student performance, contributing to improved teaching strategies and academic

achievement.

ACKNOWLEDGMENT

This work is supported by the project titled “Automated University JSU/JSI Table Generation for KUIPs,” under

the Short-Term Grant (STG) : STG-088/2023, Kolej Universiti Islam Perlis, Malaysia. The authors would also

Page 94

www.rsisinternational.org

International Conference on Islamic Contemporary Issues and Management 2025

International Journal of Research and Innovation in Social Science (IJRISS) | ISSN: 2454-6186

DOI: 10.47772/I0JRISS | ICICM 2025 - Conference Proceedings | Volume IX Issue XXIX November 2025

like to thank the reviewers and the editor for their valuable comments and suggestions, which helped enhance

this paper.

REFERENCES

1. Phung, D. V., & Michell, M. (2022, March). Inside teacher assessment decision-making: From judgement

gestalts to assessment pathways. In Frontiers in Education (Vol. 7, p. 830311). Frontiers Media SA.

2. Costa, V. G., & Pedreira, C. E. (2023). Recent advances in decision trees: An updated survey. Artificial

Intelligence Review, 56(5), 4765-4800.

3. Charbuty, B., & Abdulazeez, A. (2021). Classification based on decision tree algorithm for Machine

Lclassifificationearning. Journal of Applied Science and Technology Trends, 2(01), 20-28.

4. Schildkamp, K. (2019). Data-based decision-making for school improvement: Research insights and gaps.

Educational research, 61(3), 257-273.

5. Chang, Y. (2021). [Retracted] Based on Knowledge Recognition and Using Binomial Distribution

Function to Establish a Mathematical Model of Random Selection of Test Questions in the Test Bank.

Journal of Electrical and Computer Engineering, 2021(1), 2250300.

6. Rachman, R., Alfian, M., & Yuhana, U. L. (2023, October). Classification of Illustrated Question for

Indonesian National Assessment with Deep Learning. In 2023 14th International Conference on

Information & Communication Technology and System (ICTS) (pp. 77-82). IEEE.

7. Yu, B. (2023, August). Design and Application of Objective Questions Generating Examination Paper

System Based on Random Extraction Algorithm. In 2023 International Conference on Computers,

Information Processing and Advanced Education (CIPAE) (pp. 260-264). IEEE.

8. Zhang, J., & Li, Z. (2023, March). Design and Implementation of Machine Learning Algorithm in Question

Bank System. In 2023 International Conference on Artificial Intelligence and Education (ICAIE) (pp. 35-

39). IEEE.

9. Liu, X., & Yang, C. (2022, July). Design of Intelligent Test-composition System for English Test based on

RBF Algorithm. In 2022 International Conference on Artificial Intelligence and Autonomous Robot

Systems (AIARS) (pp. 91-95). IEEE.

10. Banujan, K., Ifham, A., & Kuhaneswaran, S. (2023). Automated question classification using Bloom's

taxonomy and deep learning techniques. Asian Journal of University Education, 19(4), 1046-1062.

https://doi.org/10.33736/ajue.5894.2023.

11. Romadhony, A., Abdurohman, R., Hasmawati, & others. (2022). Primary and high school question

classification based on Bloom's Taxonomy. 2022 10th International Conference on Information and

Communication Technology (ICoICT), 144-149. IEEE.

https://doi.org/10.1109/ICoICT55022.2022.9914842.

12. Edmodo. (September, 2008). Edmodo: Where learning happens. Edmodo. https://www.edmodo.com/

13. ExamSoft. (2019). ExamSoft: Assessment solutions for educational institutions. ExamSoft.

https://www.examsoft.com/

14. Questionmark. (2007). Questionmark: Assessment management solutions.

https://www.questionmark.com/