UTAR Institutional Repository (Universiti Tunku Abdul Rahman)
Not a member yet
6132 research outputs found
Sort by
Detecting online test cheating through user behavior monitoring
Get PDFThe shift to online testing is becoming a significant trend in the modern learning, yet this transition presents serious challenges to academic integrity and credibility of institutional qualifications. In a traditional test environment, physical supervision like physical examination setup in grand hall and attendance is a must, effectively detect cheating, but such monitoring is not feasible for remote exams, creating opportunities for students to engage in dishonest behaviour. The proposed system, "EyeGuard," aims to assist and solve this issue by employing a computer vision-based eye gaze detection system that uses a student webcam to track their eye movements during an online test. By analysing eye gaze patterns and browser incident to identify inappropriate activity, such as looking at unauthorized materials during the test and switching the tab, the system can detect potential cheating in real-time, allowing any suspicious behaviour to be reported instantly for necessary investigation. The principal objective of this project is to provide a reliable and effective solution for monitoring online exams that reduces the reliance on human supervisors, which can be costly and impractical at scale. Through the automated detection of suspicious behaviour, "EyeGuard" fosters a more confident and fairer environment for online test, ensuring the integrity of the examination process while offering a scalable, and low-cost solution for educational institutions
TagT: A versatile ANPR solution for diverse applications
Get PDFTraditional Automatic Number Plate Recognition (ANPR) systems, which focus solely on
license plate numbers detection and recognition are vulnerable to fraud. This project presents
the design and implementation of TagT, an advanced ANPR framework that enhances security
through multi-attribute car recognition. TagT integrates three key components: a YOLO11n
model for high-speed car detection, a ResNet18 model with cosine similarity for intelligent
frame optimization and the Gemini model for robust recognition of a car's license plate number,
brand and colour. An extensive preliminary investigation justifies the selection of these models
over numerous alternatives. The final, implemented system features a native iOS application
and a Python back-end. A comprehensive evaluation was conducted to validate the prototype's
performance, focusing on two key areas: Efficiency and Accuracy. The evaluation of the
architecture's efficiency demonstrated a 92.4% reduction in frames sent for analysis, which
resulted in a 91.9% decrease in API costs and an 87.6% decrease in API latency compared
to a baseline approach. Furthermore, the system's real-world accuracy was validated across 160
demanding tests in varied conditions, achieving an average Overall Success Rate of 83.75%
and a near-perfect Car Brand Accuracy of 98.13%. Overall, TagT provides a versatile, costeffective
and scalable solution that successfully addresses the limitations of traditional ANPR,
enhancing public safety and car management
The impacts of social media use on English language usage among private university students: The mediating effect of intrinsic and extrinsic motivation
Get PDFConcerns have arisen regrading social media’s influence on students’ language practices with the widespread use of social media. This research examines the relationship between social media use and English language usage among Malaysian private university students, with intrinsic and extrinsic motivation as the mediating factors. The study also employed a quantitative and crosssectional design guided by Self-Determination Theory (SDT) and Technology Acceptance Model (TAM). Convenience sampling was used to recruit a total of 321 students from various faculties. Data were collected through an online survey questionnaire and analysed using SPSS for data cleaning and SmartPLS for Partial Least Squares Structural Equation Modeling (PLS-SEM). Findings revealed that social media use significantly impacted English language usage, with intrinsic motivation emerging as a strong and significant mediator. In contrast, extrinsic motivation demonstrated an insignificant mediating factor. The findings indicated the importance of promoting intrinsic motivation in language learning and offer practical and theoretical implications for employing social media to improve English proficiency in higher education
Negative impacts of the use of artificial intelligence in English language learning among university students and precautionary measures to mitigate
Get PDFThe emerging use of Artificial Intelligence (AI) in higher education raises concerns regarding the potential negative effects that may result from an extensive AI usage in learning English. This research aims to examine the negative impacts of long-term AI usage on English language learning, and identify the precautionary measures to reduce such adversities. This paper implements a mixed-methods approach, whereby a questionnaire consisting of 5-point Likert scale questions and open-ended questions is distributed to 150 undergraduates majoring in English to obtain their perspectives on the research topic. The data collected are analysed via descriptive analysis and thematic analysis. The main findings demonstrate that Technology Dependence and False Information are two of the most prominent adverse effects of using AI in learning English. Additionally, the findings also depict eight themes derived from the participants’ responses regarding the precautionary measures, which encompass areas such as practical tasks, AI literacy, AI ethics, plagiarism, assessment modification, and cognitive skills. The main findings emphasise the importance of raising public’s awareness on the potential negative consequences in using AI to learn English as well as the mitigation strategies to build a well-balanced AI-integrated teaching and learning environment
DOI and ISBN citation converter
Get PDFThis project developed a web-based citation generator to automate the accurate formatting of academic references, addressing a significant need among students and researchers who struggle with manual citation processes. The system supports both Digital Object Identifiers (DOIs) and International Standard Book Numbers (ISBNs), generating citations in multiple styles including APA, Harvard, and IEEE. The methodology adopted is an Iterative approach. The application employs a three-tier architecture consisting of a lightweight HTML/CSS/JavaScript front end, a high-performance Python backend built on FastAPI, and DuckDB for persistent metadata storage. The system fetches metadata from external APIs asynchronously, using CrossRef for DOIs and Open Library for ISBNs, and using Pydantic models to provide strict input validation. A key innovation involves using DuckDB as a local cache to minimise redundant API calls and significantly improve response times. Testing of the system demonstrated its reliability in generating accurate citations, managing invalid inputs with ease, and efficiently processing multiple references through its advanced bulk upload functionality. The project provides a practical and user-friendly solution that saves time while maintaining academic integrity by reducing formatting errors. Future developments may include additional citation styles, export features, and cloud-based deployment to enhance scalability.
Keywords: citation, APA style, IEEE style, Harvard style, citation converter, web application
Subject Area: PN172 – Literary Composition Technique
Web based automated e-invoicing system
No full textThis project focuses on the development of an e‑Invoice system designed to modernize and automate invoice generation, submission, and verification in compliance with regulatory standards. The purpose of this project is to address the challenges in manual invoicing, including manual errors, lack of transparency in manual invoicing, and the increasing number of fraud risks that had been a pervasive problem within the domain of financial technology and digital tax compliance. The existing problems motivating this project stems from the need for a secure, auditable, and standardized process for generating and submitting invoices, which in turn, ensures data integrity and legal compliance. To address this, a modular approach was adopted which involves mapping invoice data to the UBL 2.1 JSON standard, performing canonicalization for deterministic representation, and cryptographically signing it before submission. Besides, the system integrates with LHDN tax authority APIs for asynchronous validation, generates QR codes for verification, supports cancellation of invoices, notifications, and maintains a ledger for reporting and reconciliation. The preliminary results demonstrate that the platform reliably produces compliant e-invoices, which also accurately detect data mismatches, and enables taxpayers to track invoice status in real time, while QR code verification enhances transparency for end users. The significance of this research lies in its ability to streamline the invoicing process, reduce human error, and provide an auditable digital record, contributing to greater efficiency and regulatory compliance. Therefore, this project has demonstrated a scalable framework, which includes document signing, submission automation, and real-time verification in financial technology applications.
Keywords: e-Invoice, LHDN MyInvois Integration, RESTful API, Web Application Development, API Integration, OAuth2 Authentication, Digital Signature, Tax Compliance.
Subject Area: QA76.75-76.765 Computer softwar
Item-level machine learning approach to identify influential predictors in self-report mental health scales
No full textThis research introduces a innovative Long to Short approach on the DASS-42 mental health assessment tool for assessing stress levels among adults using machine learning. The data first retrieved for the mental heath assessment from Kaggle. The sum of the scores then obtained based on participants’ answers to every items in the complete questionairre. Next, feature selection techniques were applied to identify a selected items from the assessment based on participants’ responses, aiming to accurately predict outcome. Machine learning models were trained to get the smallest set of items required to reach a prediction accuracy of 95%. This study found that just three items are sufficient to predict stress status with at least 95 % accuracy compared to the full-scale assessment, using XGBoost and MLP model. However, demographic data such as age, gender, education level, and cultural background were not included in the analysis. The exclusion of these variables may limit the generalizability of the results, as demographic factors can influence how individuals respond to psychological assessments.
Keywords: machine learning, DASS-42, stress assessment, feature selection, MLP, mental health screening
Subject Area: QA76.75-76.76
槟榔屿广福宫碑文与周边华人社群之关系探析 : An analysis of the relationship between the inscriptions of Kong Hock Keong Temple in Penang and the surrounding Chinese community.
No full text本文以槟榔屿广福宫及其碑文为核心研究对象，探讨十八世纪末至十九世纪槟 榔屿社会的形成及华人移民的历史进程。文章首先梳理了英国东印度公司入驻 后槟榔屿从荒岛到重要贸易港口的转变，分析自由港政策与地理位置优势如何 促进国际商贸发展，并吸引大批华人移民下南洋定居，推动当地经济与多元文 化形成。其次，广福宫的建立不仅满足了移民的宗教信仰需求，还在早期承担 协调社群事务和调解纠纷的功能，体现了宗教场所在殖民地社会治理中的独特 作用。本文对碑文的细读揭示了闽粤籍群体在经济层面及社会地位上的差异， 也反映了商业资本对宗教建设的推动作用。碑文修辞显示华人逐渐将槟榔屿视 为“第二故乡”，折射其身份认同的转变。整体而言，本研究通过宗教建筑与碑 刻史料，呈现华人移民在槟榔屿社会发展中的重要角色，并为理解海外华人文 化传承与社区建构提供新视角。 【关键词】槟榔屿、广福宫、碑文研究、华人移
Examining the mediation effect of trust and functional barriers in influencing purchase intention via online travel agencies
No full textI would like to express my deepest gratitude to everyone who supported me throughout my research journey.
First and foremost, my sincere appreciation goes to my research supervisor, Dr. Mandy Yeong Wai Mun. Her unwavering support, valuable guidance, and insightful ideas have been instrumental in the successful completion of this project. Dr. Mandy’s enthusiasm for International Business and her extensive experience provided me with a wealth of knowledge and a clearer understanding of the subject. Her dedication and encouragement helped me navigate the challenges of the
research process, for which I am truly grateful.
I also extend my heartfelt thanks to the respondents who generously participated in my research questionnaire, despite their busy schedules and some challenges with
Google Forms. Their willingness to contribute has provided invaluable insights for this study, and I sincerely appreciate their time and effort.
Furthermore, I am profoundly grateful to my friends and family for their continuous support. My friends actively sought out suitable respondents for my study and stood
by me during challenging moments, understanding my concerns. My family’s emotional encouragement and care have been crucial in keeping me motivated throughout this journey.
Lastly, I would like to express my gratitude to Universiti Tunku Abdul Rahman (UTAR) for the opportunity to undertake this Final Year Project. The university's resources, including teaching materials and research facilities, have greatly supported my work and contributed to the successful completion of this study
Portable electrochemical impedance spectroscopy measurement device for battery quality check
Get PDFThis project presents a portable Electrochemical Impedance Spectroscopy (EIS) device
for real-time lithium-ion battery (LIB) health assessment, addressing the accuracy, cost,
and deployability limits of conventional methods such as Coulomb counting and opencircuit
voltage analysis. The prototype integrates a miniaturized impedance front end
with an embedded controller, delivering 19-point frequency sweeps and 1 kHz sampling
while rendering live Nyquist plots on a touch display and logging results to microSD
in CSV format. The system emphasizes practical field use: auto-scaled visualization,
non-blocking timing for looped measurements, and a queue-based logger that maintains
responsiveness under sustained writes. Repeatability tests show stable overlays for the
same cell and clear separation between different cells, indicating reliable acquisition
for downstream analytics. In parallel, the design establishes an AI-ready pipeline by
structuring data for feature extraction and model development to refine State of Charge
(SoC) and State of Health (SoH) estimation. Preliminary analysis supports the
feasibility of on-device or near-edge inference using lightweight models to enable
predictive maintenance, early degradation detection, and optimized charging strategies.
By combining portability, real-time operation, and data-driven diagnostics, the
proposed EIS platform offers a cost-effective path toward proactive battery
management in electric vehicles, renewable energy storage, and consumer electronics