UTAR Institutional Repository (Universiti Tunku Abdul Rahman)
Not a member yet
6132 research outputs found
Sort by
Factors influencing the adoption of AI-driven chatbots for mental health support among university students in Malaysia
Get PDFThis study investigates the factors influencing the adoption of AI-driven chatbots for mental health support among university students in Malaysia, using the Unified Theory of Acceptance and Use of Technology 2 (UTAUT2) as the guiding framework. In particular, the efficacy and efficacy of the performance expectancy (PE) and social influence (SI), facilitating condition (FC), and habit (HT) on adopting AI-driven chatbots as a mental support mechanism in universities in Malaysia is investigated. Quantitative methodology was employed, whereby the researchers collected 391 valid reactions of an online survey launched among students in the universities of Malaysia. Data were analysed using SPSS, with Pearson Correlation and Multiple Regression Analysis applied to assess the relationships between variables. The results indicate that performance expectancy (PE), social influence (SI), and habit (HT) are significant factors, and the facilitating condition (FC) is insignificant in the adoption of AI-driven chatbots to support the mental health of university students in Malaysia. This knowledge has practical recommendations that could be offered to universities, developers, and policymakers who intend to roll out student support services with AI-powered mental health tools. The research then ends with future research options and strategies to be used to increase adoption in various educational backgrounds. Keywords: AI-driven chatbots, mental health support, university students, adoption intention, UTAUT2, performance expectancy, social influence, facilitating conditions, habit, Malaysia Subject Area: HM1176-1281 Social influence. Social pressure, T1-995 Technology (General
Supermarket shopping assistant for the visually impairs
Get PDFThis project introduces a Supermarket Shopping Assistant designed to support Visually Impaired (VI) individuals in performing grocery shopping independently. Shopping in supermarkets presents major challenges for VI users, as identifying products, locating items on shelves, and reading packaging information typically require external assistance. To address this gap, the proposed system integrates computer vision, optical character recognition, and text-to-speech technologies into an Android mobile application that provides real-time product recognition, hand guidance, and product information retrieval.
The system architecture was developed around two complementary approaches for product identification: a product detection model using Roboflow API, and a text detection approach leveraging Google Cloud Vision API. While the detection model achieved high accuracy in recognizing standard grocery products, the text detection approach provided greater robustness by always generating an output, even when object detection failed. In addition, hand detection was implemented using MediaPipe to guide the user’s hand toward the detected product through audio feedback. Once the item was acquired, the user could capture its packaging details, which were processed using OCR and summarized by the Gemini API before being read aloud, ensuring clarity and conciseness.
Evaluation results demonstrated that the system successfully fulfilled its objectives, with each module performing reliably under controlled conditions. Although challenges such as dependence on cloud services were noted, the system still provided a practical and accessible solution. By enabling VI individuals to identify, reach, and evaluate grocery products independently, this project contributes to enhancing daily autonomy and highlights the potential of AI-driven assistive technologies to promote inclusivity in society
A multi-agent rag system for auditable token distribution
Get PDFThis project develops a blockchain-based subsidy program delivered as a web application
to enhance the efficiency and transparency of government subsidy distribution. By
leveraging blockchain technology, the system ensures secure and auditable allocation of
funds, thereby strengthening public trust in subsidy mechanisms.
The research methodology involves a review of existing blockchain applications in the
public sector, followed by phased development using Visual Studio Code, Supabase, and
the Ethereum Sepolia Network. The novelty of this work lies in its integration of multiple
innovations. First, an ERC-20 token, MMYRC (Mock Malaysia Ringgit Coin), is created
and airdropped to eligible citizens to demonstrate that the entire distribution process can
be conducted on-chain. Second, eligibility is determined through a rule-based scoring
matrix supported by Retrieval-Augmented Generation (RAG), providing decisions that are
both transparent and explainable from government datasets. Third, lightweight AI agents
(smolagents) implement a dual-analysis framework that combines the RAG-based
approach with a formula-driven burden score, allowing for systematic comparison between
interpretability and flexibility in eligibility determination. Finally, zero-knowledge proofs
(ZKPs) are explored as a privacy-preserving mechanism, enabling citizens to prove their
income bracket (B40, M40, T20) without disclosing precise income data. A mock trusted
setup with the Inland Revenue Board of Malaysia is used to verify digitally signed income
information, ensuring authenticity while preserving privacy.
The expected outcome is a fully functional prototype web application that reduces
administrative overhead, improves the auditability of subsidy allocation, and enhances the
effectiveness and public legitimacy of government subsidy programs
Arfict: Indoor navigation mobile application using augmented reality (AR) for FICT
Get PDFAs commercial and educational institutions grow more complex, indoor navigation has
become essential. Traditional GPS navigation, effective outdoors, struggles indoors due
to signal interference from walls and ceilings. This is even more significant in large,
multi-floor buildings such as university campuses, where GPS signals are blocked.
Additionally, 2D maps are not capable of providing real-time, adaptive directions,
leading to inefficiency and user frustration. Hence, the ARFICT project aims to develop
an Augmented Reality (AR)-based indoor navigation system for the FICT building at
UTAR, providing a real-time, interactive navigation experience that enhances
wayfinding in complex indoor environments. The project's contribution goes beyond
UTAR, forming a basis for AR-based navigation in various large indoor environments,
such as shopping malls, airports, and hospitals. The use of external hardware, like
Bluetooth beacons, is eliminated, thus offering an affordable solution for navigation in
multi-floor buildings. The system employs a marker-based localisation method using
QR codes decoded with the ZXing library to establish the user’s initial position.
Navigation paths are optimised using the A* algorithm, while ARCore integrates sensor
data from the device’s gyroscope and accelerometer to enhance tracking accuracy.
Users are guided with AR-based path visualisation, distance indicators, and voice
instructions, ensuring intuitive and efficient navigation across multiple floors. In
summary, this project successfully developed an AR indoor navigation application to
provide advanced indoor navigation, improve efficiency, and provide a reliable
wayfinding for students, staff, and visitors in the multi-floor FICT building, enhancing
the user experience in complex indoor environments
Video surveillance: explosion detection
Get PDFThe proliferation of surveillance technologies has emphasised their pivotal role in
enhancing public safety by monitoring and detecting anomalies in real time. Among
the anomalies, explosions present a grave threat due to the potentially resulting major
loss of life, widespread panic and significant destruction of property. However,
traditional surveillance systems are limited by their reliance on human monitoring,
which is susceptible to oversight due to fatigue or distractions. Thus, this research
focusses on developing an intelligent explosion detection surveillance system that is
capable of early and accurate explosion detection. Explosions typically happen in a very
short timeframe, often just a few seconds, leading to significant challenges for the rapid
and accurate identification of explosions' unique visual patterns immediately. Hence,
this study proposes to embed computer vision and advanced image processing
algorithms, such as Motion History Images (MHI) and Motion Energy Images (MEI),
into the intelligent explosion detection video surveillance system. By leveraging three
motion-based variables, including motion ratio, new pixel ratio and optical flow values,
together with three detection approaches, namely global detection, non-eroded
detection and eroded detection, the system demonstrates the effectiveness of motion
based methods in detecting explosions at an early stage with acceptable performance.
Eventually, this approach aims to minimise explosions' damage by enabling immediate
responses, preventing the spread of fires and the occurrence of secondary explosions
Factors influencing passengers' satisfaction with public transportation services quality in Klang Valley
Get PDFIn crowded cities like Malaysia's Klang Valley, public transport is important in daily life. As the population grows, then worsen traffic congestion issue, improving public transportation services has become necessity, to make travel easier and more convenient. Tangibility, reliability, responsiveness, assurance, and empathy are the five main characteristics that will be examined in this study as it attempts to determine the elements that affect passengers' satisfaction with Klang Valley public transport services. Data was gathered from passengers who frequently take public transit in the area using a questionnaire. SPSS 30.0 was used to evaluate the results. The results show there is a substantial connection between passenger pleasure and four factors: tangibility, reliability, assurance, and empathy. This means that passengers value clean and comfortable facilities, on-time and dependable services, feeling safe while traveling, and staff who show care and respect. However, responsiveness, which refers to how quickly and helpfully staff respond to passengers’ needs, did not have a strong influence on satisfaction. The study mentioned the important of focusing on the areas that truly matter to passengers. In conclusion, understanding what passengers care about most can lead to better service quality, higher satisfaction, and increased public transport usage in the future. Keywords: Service Quality Model (SERVQUAL), Passenger Satisfaction, Light rail transit (LRT), Public Transport, Urban Transport Subject Area: HE305-311 Urban transportatio
Integrating ChatGPT chatbot into a hospital website
Get PDFThis project presents the development and integration of a ChatGPT-based chatbot
system and a chatbot management system into the UTAR Hospital website to improve
user interactions, especially for patients, visitors, and hospital staff. As most
healthcare platforms begin to get covered by AI, this chatbot system has been
designed and developed in such a way to make the dissemination of information
easier, provide accurate healthcare guidance, and lighten the workload for healthcare
workers by reducing administrative work. By customizing the behavior of ChatGPT
through domain-specific prompt engineering, the chatbot delivers real-time responses
to common hospital-related queries without requiring fine-tuning the model. A userfriendly interface has been developed and integrated into the UTAR Hospital website
to improve accessibility, user experience, and operational efficiency. The chatbot
system was further improved with a management interface that allows hospital
administrators to oversee live conversations, customize chatbot responses, manage
datasets, adjust the chatbox appearance, and control chatbot behaviour through rolebased access. Moreover, system testing confirmed that the chatbot performed reliably
in providing multilingual replies, accurate doctor and service information, and
effective admin monitoring. This combination reduces administrative workload,
improves user experience, and supports UTAR Hospital’s digital transformation
efforts
Effect of different tea types (Camellia sinensis) on antioxidant and sensory properties of kombucha fermentation
Get PDFKombucha is a functional fermented tea beverage that has been gaining consumer
interest due to its rich bioactive compounds. While green tea kombucha and black tea kombucha are well studied, limited research has focused on kombucha fermented with oolong tea (semi-oxidised) and dark tea (post-fermented), which
have different phytochemical and flavour profiles. The study found that the types
of tea, fermentation time, and their interaction significantly affected the total phenolic content (TPC), total flavonoid content (TFC), and antioxidant activities (DPPH and FRAP) of kombucha. Green tea kombucha (GTK) consistently exhibited the highest TPC (up to 788.48 �} 5.84 mg GAE/L), TFC (up to 420.30 �} 2.92 mg QE/L), and antioxidant activities (up to 73.15 �} 0.38% in DPPH inhibition, 16.19 �} 0.07 mmol Fe2⁺/L in FRAP), followed by oolong tea kombucha (OTK) and dark tea kombucha (DTK). All samples exhibited a significant increase in phytochemical content and antioxidant activities by Day 7, followed by a decrease by Day 14. From Day 0 to Day 7, GTK had the highest increase in TFC (+159.39 mg QE/L) and FRAP (+4.44 mmol Fe2⁺/L), while the lowest increase was in DTK (+65.75 mg QE/L, +1.71 mmol Fe2⁺/L). From Day 7 to Day 14, the highest decreases in phytochemical content (–85.45 mg GAE/L, –53.03 mg QE/L) were observed in GTK, while the highest decreases in antioxidant activities (–4.24 %, –2.12 mmol Fe2⁺/L) were observed in OTK. In the sensory evaluation, GTK and OTK received significantly higher ratings for aroma, sourness, and overall acceptability than DTK, with no significant difference in colour and sweetness among all types of kombucha. Overall, GTK was more recommended due to its higher phytochemical content, antioxidant activities, and consumer acceptability
Bursa stock recommender using technical analysis
Get PDFInvesting in the stock market can be an effective strategy for wealth creation and financial security. However, current stock recommendation systems often lack user-centric features such as personalized insights and effective communication channels for investors. These limitations become challenges for novice investors in navigating financial data and making informed decisions. This project introduces Bursa Stock Recommender, a stock recommendation application that is designed to simplify the process of selecting a stock through financial performance analysis and technical indicators. This application tailored investment recommendations by leveraging historical financial data and other important metrics such as cash flow, price-to-earnings ratio, and dividend yield. It also incorporates advance charting tools for expert investors while maintaining the simplicity of this application for beginners through a user-friendly interface and predefined screens. A key feature of this application is the inclusion of a collaborative communication platform, which enables investors to share insights, discuss strategies, and exchange ideas. Additional functions such as risk management tools, customizable watchlists, and alert systems are also provided to enhance decision-making capabilities. In short, the core value of this project is to promote broader participation in the stock market and empower investors to reach their financial goals
Flood management system: geo based information and monitoring module
Get PDFFloods stand as some of the most devastating natural disasters because they provoke enormous damage, the loss of human lives, and the disruption of communities. Despite the advancement of flood management systems, most of the existing solutions experience problems regarding real-time flood information, inefficient relief center coordination, and ineffective evacuation processes, which delay responses and poor resource allocation. This project introduces FloodGuard, which a comprehensive flood management system application to address the existing gaps and enhance real-time disaster response. One of the key novel features is the implementation of an automated relief center assignment algorithm, which dynamically assigns the user to a nearby relief center based on the risk assessment, proximity and capacity rather than manually searching for relevant centers. This system involved a centralized command center for administrator to review and approve flood reports, monitor real-time flood summaries and user records for better decision-making. Moreover, FloodGuard employs an innovative approach to human-scale water level measurement wherein users input their height and mark the water level on their body to generate more accurate flood depth data. Then the data feeds into a multi-dimensional risk assessment framework which calculates user priority scores based upon demographic factors, building characteristics, and real-time exposure metrics. Integrated with dynamic analysis of flood severity, the system incorporates automated risk-prioritized relief center assignment. Other features include push notifications for residents within flood zones, route guidance to centers, flood reporting and structured emergency communication systems that provide users with necessary services such as medical aid. In addition, all the submitted reports are verified by the administrator before being published to ensure reliability. Developed in React Native, Expo, and Firebase, FloodGuard demonstrates significant improvements in disaster response efficiency and community preparedness, enhancing safety and resilience from floods