MST-Department of Computing & Information Technology

Permanent URI for this collection

https://repository.ku.ac.ke/handle/123456789/9331

Browse

Now showing 1 - 8 of 8

Customers’ Perception of Mobile Banking and Financial Performance of Commercial Banks in Nairobi City, Kenya
(Kenyatta University, 2023-11) Munyasia, Nelima Rose; Stephen Titus Waithaka
A Hybrid Model for Text Summarization using Natural Language Processing
(Kenyatta University, 2022) Karanja, James Mugi; Eliud E.O. Obere
Small and medium sized businesses are key aspects of economic progress in every country and their internationalization is thought to be essential to the growth of the economy and firm growth. Micro, small, and medium-sized businesses have endured a great deal of failure and death in Kenya, despite the fact that they are crucial for increasing employment, capital base, and revenue. The performance of the small and medium sized enterprises in Kenya’s internationalization remains dismal because more focus by the government has been on foreign investors. This study sought to establish the key internationalization strategies that impact Micro and SMEs performance in Nairobi City County. The general objective of the study was to investigate effects of internationalization strategies and performance of micro, small and medium sized manufacturing enterprises in Nairobi City County. The specific objectives of this study therefore were to determine the effect of managerial competence, network structure, operating network and international market knowledge on performance of Micro, Small and Medium size Manufacturing Enterprises in Nairobi City County, Kenya. The study was anchored on study: Stakeholders’ Theory, Tradition Foreign Direct Investment Theory, Stage Model Theory and The Network Approach. This study used a descriptive survey design. The target population was 262 exporting micro small and medium sized manufacturing enterprises that are located in Nairobi City, County. A sample size of 53 potential respondents were randomly selected representing 20% of the total population as justified under sampling technique. The key source of data was primary, obtained using structured questionnaires whose reliability and validity were ascertained. Descriptive and inferential analysis were the two methods that were utilized. The study found that managerial competence, operating environment and international marketing knowledge had significant effect on MSMEs performance. Managerial competence is thought to be an important determinant of internationalization for MSMEs. The study recommended global partnership to enhance global experience and innovation capability on SMEs' export success.
Automated Examination Generation using Natural Language Processing and Artificial Neural Network
(Kenyatta University, 2023) Ndirangu, Peter Ndegwa; Elizaphan Maina
The examination process is a key activity in evaluating what the learner has gained from the study. Institutions of Higher Learning (IHL) perform the activity by administering tests which comprises of questions and answers. Cognitive level, weight of the question, and topic coverage are key factors to consider when setting exams. The world today has largely focused on the automation of exam generation which has been ongoing with dire need during the period of the Covid-19 pandemic when education was greatly affected, leading to embracing online learning and examination. The process has taken shape; however, the automation process can be improved by incorporating machine learning algorithms in the process of setting examination. In view of this, the project focused on implementation of a question classification model that uses Neural-Network algorithm (NN) and Natural Language Processing (NLP) to determine questions cognitive levels based on the revised Bloom's Taxonomy. The iterative method of software development was adopted to provide room for continuous improvement. The developed model was put under test with a couple of questions obtained online. The effectiveness of the model was determined by subjecting it into database of 600 questions resulting to an accuracy of about 71%. An Application Programming Interface (API) and Moodle Learning Management System (LMS) plugin were consequently developed to allow integration of the model with an existing system. The deep learning approach was applied to predict cognitive levels of questions based on Bloom’s taxonomy and the resulting questions were made available to the instructor through the LMS interface. Future research should focus on the use of convolutional reinforcement learning to establish its effectiveness in question classification as well as perform comparison with various algorithms.
Improving Effectiveness of Industrial Placement Experience Using a Recommender System
(Kenyatta University, 2023) Ndolo, Daniel Mulinge; J. Kandiri
Different classifier systems have been developed as a result of technological advancement to replace traditional job-search techniques. Finding locations for their industrial attachments is difficult for students in tertiary institutions. By making this training option, they are restricted to open elements like geographic coverage and a limited understanding of industry players. This study used mixed research methodology to conduct its investigation which included survey research and software development methodologies. A multi-criteria classifier called PlacementKe was developed to create user profiles that are used to forecast and provide recommendations while matching a student with the right company for their industrial attachment training. The classifier was based on the hybrid collaborative filtering algorithm. The recommendations generated were anchored on the user profiled interest and weighted ratings. Using a preexisting data set produced by Kaggle, a base model was created using a pre-trained base model. The base model’s accuracy level was 92%. The base model was used to create parameters that were applied in the development of a custom placement recommender system. The model was implemented to a prototype application, which was evaluated using actual users and data. The user comments were recorded on a survey form and examined. According to the analysis, 77.78% of users were happy with the system’s overall performance. The inquiry used in the literature evaluation revealed a need for automation in industrial attachment placements to increase their efficiency and speed. The qualitative and quantitative objectives that guided the system design were both met by the research design. The system designed achieved a high user satisfaction rate. This study suggests that the system be developed and implemented for use in higher education institutions. Future work extensions are also discussed.
Enhancing Asset Security by Integrating Internet of Things on Non-Powered Assets
(Kenyatta University, 2023) Mwema, Joshua Mueke; John Kandiri
Advancements in the field of asset security have emerged as a result of the Internet of Things (IoT)’s explosive expansion. Several innovations have featured systems ranging from smart home automation to asset tracking and monitoring. The systems use different technologies such as Radio Frequency Identification (RFID), Global System for Mobile communications (GSM), General Packet Radio Services (GPRS), Wireless Fidelity (Wi-Fi) among others to secure the assets. As technology improve, intruders also update themselves with intrusion skills and knowledge, and this has led to the emergence of more sophisticated challenges in terms of asset security. To curb these challenges, this project proposes an asset security system that will be comprised of IoT integration, real-time alerts, and power autonomy for remote monitoring of the non-powered assets. Besides, a tamper-proof unit will be used to detect when the asset has been broken into, which will prevent intruders from compromising the security system inside the asset. In that case, a notification will be sent to the asset owner through the GSM SMS functionality. Further, based on the distance from the pre-set distance, the device will map the value to asset security states namely; zero to 0.5, 0.5 to 1 and greater than 1 where they will be interpreted as low, medium, and high-level security threat states respectively. This research project will bring on-board geofencing and remote-control capabilities to asset security systems so that the location of a device can be tracked when it is relocated.
An Intelligent Based System for Supporting Personalised E-Learning
(kenyatta university, 2023) Kivuva, Faith Ngami; Elizaphan Maina
Most traditional e-learning systems fail to provide the intelligence to guide a learner according to their learning style. However, intelligent agents can be created to perform the role of guide to a student depending on a predetermined learning style. In view of this, the study discusses how to design, develop and implement intelligent agents for supporting personalized e-learning based on a predetermined learning style. The main objective of this study was to design and implement an intelligent e-learning system based on intelligent agents for supporting personalized e-learning. The system, which is based on intelligent agents, provides some intelligence and supports dynamic learning. Each learner has different levels of achievement depending on their learning styles and gets personalized feedback/recommendations. Three intelligent agents were developed; a learner agent, a tutor agent, and an information agent. The learner agent, which has an AI engine, uses deep neural networks to provide a recommendation to the learners based on their learning styles. The tutor agent accesses what the learner has accessed and passes this information to the learner agent which then recommends the appropriate materials. The information agent presents the recommendations/feedback of the learners through the Moodle user interface. The learning styles of the students are determined by filling out a Visual, Aural, Read/Write, and Kinesthetic (VARK) questionnaire. The three agents were developed using the Prometheus methodology. They were also tested and integrated into Moodle Learning Management System (LMS). This integration allows learners who are using LMS such as Moodle to learn based on their learning style. The results indicate that it is possible to train a learner agent using deep neural networks and provide personalized learning to the learner based on the learning style. Future studies need to focus on using data collected in a learning management system to identify learner styles instead of using the VARK questionnaire. Additionally, it is necessary to use other learning styles models, such as the Filder-Silverman model, and the Kolb learning style model among others, to identify learning styles and conduct an experimental study to determine their effectiveness in personalized learning with intelligent agents.
Intelligent Conversational Agent for Enhancement of Online Communication in Selected Universities Using Pattern Matching Algorithm
(Kenyatta University, 2022) Kuria, Isaac Njoroge; Harrison Njoroge
University websites and online portals are the primary means through which potential students and other stakeholders find important information about an institution. University websites are essential to these organizations’ marketing and communication efforts. In websites, a lot of information is spread across numerous amounts of web pages. Navigating through these web pages to locate relevant results, according to the user needs, can be non-resulting, time consuming and annoying at times. There is need to complement these websites with the use of an AI Chatbot (UniBot) in order to serve more efficiently. To address this problem, the research project proposes to design, develop and implement such an agent that will engage online users of universities websites and online channels efficiently and in real time. The project initially aims at performing an extensive literature survey on intelligent conversational agents and the feasibility of applying them in enhancing online communication in universities. This will guide the design development, implementation and testing. The project shall utilize an iterative - incremental methodology to aid in design and development of UniBot, using AIML (Artificial Intelligent Markup Language) Pattern matching algorithm on the Pandorabot (AIAAS) platform. This will generate high quality training data, with which, the agents Natural Language Understanding (NLU) model can be trained. This will make the agent (UniBot) capable of handling user requests efficiently at run time. The agent will be integrated to the university website by use of an API. Finally there will be a provision to train and test the agent using data which will be made available by Online Communication/ University Website department at Kenyatta University.
Application of mobile phone in crime prevention within Central division, Nairobi city County
(Kenyatta University, 2017-08) Wambugu, Paul M.
ABSTRACT The study sought to establish application of mobile phones applications by police officers in crime prevention in Central Police Division, Nairobi City County. Application of mobile phone in the police service is underutilized making the organization not to fully benefit from its usage. Understanding use of mobile phone applications by the police officers may assist the organization in crime prevention efforts. Due to its ability to engage consumers in a timely and direct manner at low costs, mobile phone applications are relevant for the police organization. The study was guided by the Mobile Technology Acceptance Model (M-TAM) as the success of the mobile phone implementation depends on perceived usefulness (PU) and perceived ease of use (PEOU).The targeted population in this study was police officers serving at Central Division of Nairobi County (782). A sample of 155 police officers was recruited using random stratified sampling procedure for this study. Questionnaires were used for data collection for both qualitative and quantitative data. Qualitative data was coded and analysed thematically after interpretation of theme while descriptive data was analysed using descriptive statistics by use of Statistical Package for Social Sciences (SPSS version 20.0). Mobile phone was found to be effective tool in prevention of crime in Nairobi Central. Mobile phone applications such as Facebook, Whatsapp, Twitter, emails and short text messages were found through adapted technology acceptance model (TAM) as to effectively contribute to crime prevention based on respondents’ perspective. Police effort to use mobile technology was found to have significant effects on crime prevention. Based on these findings, the study recommends development of customised police mobile phone applications and enhancement of applications use through capacity building among police officers.

Browse

Recent Submissions