Smart Computing and Systems Engineering - 2023 (SCSE 2023)

Permanent URI for this collectionhttp://repository.kln.ac.lk/handle/123456789/27032

Browse

Now showing 1 - 20 of 56

Process Improvement Framework for DevOps Adoption in Software Development
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2023) Jayakody, J.A.V.M.K.; Wijayanayake, W. M. J. I.
DevOps is welcomed by software development companies in recent years as a novel approach attached to the Agile software development methodology. Yet, they are in trouble with implementing DevOps because it doesn't just concentrate on technological changes. It alters the software development process more broadly. To assist this challenging process, DevOps maturity models have been established by a few scholars in recent years. Nevertheless, those models consist variety of drawbacks as; the majority of them have not been properly evaluated and published. This research aimed to provide a critical evaluation of the data available in existing studies on the DevOps maturity models and to propose a DevOps adoption process improvement framework that is validated by industry practitioners. To accomplish this target, a systematic literature review was applied and studied the available DevOps maturity models, weaknesses, and strengths of those models. A new framework for DevOps process improvement is developed by monitoring and contrasting the available data. Furthermore, it was assessed by an interview survey to strengthen the research's overall goal. The study presents a verified DevOps process improvement model which consists of four main DevOps success areas; DevOps practices, DevOps team, DevOps culture, and DevOps measurement. Each area follows five maturity levels starting with beginning to expert. This framework assists software development companies in obtaining benefits while reducing the difficulties associated with DevOps adoption.
Web-Based Data Hiding: A Hybrid Approach Using Steganography and Visual Cryptography
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2023) Ediriweera, Seniru; Dilhara, B.A.S.; Disanayaka, Chamara
In today's digital age, protecting sensitive data during transmission and storage is a critical concern. The rise of cyber threats has made it essential to develop secure communication channels to prevent unauthorized access and theft of confidential information. In this research, we propose a system that utilizes a combination of steganography and visual cryptography for secure data hiding. The main goal of this research is to address the issue of secure communication by concealing information in a digital image using steganography. After encoding the text in the image, the resulting steganographic image is divided into two shares using visual cryptography, ensuring that the data is protected from unauthorized access. This approach offers a practical and effective solution for secure data hiding, which can have potential applications in fields such as information security, privacy protection, and digital forensics. Overall, this research offers a viable solution to the problem of secure communication, which can help safeguard confidential information in today's digital world.
A Review of Recent Trends in Sri Lankan Social Media Analytics Research
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2023) Sandaruwani, M.D.; Hewapathirana, I.U.
Due to industry demands and massive applications, the social media landscape is rapidly expanding. However, in Sri Lanka, analyzing social media data is still considered a young research topic. This article examines the present status of social media analytics research in Sri Lanka, highlighting selected technologies and applications and discussing their proven and future benefits. The primary goal of this research is to provide information regarding social media analytics usage in Sri Lanka and to identify shortcomings in this area. We select 45 publications published between 2013 and 2022 from the most used web- based databases, including Google Scholar, IEEE Xplore, ScienceDirect, Springer, and ResearchGate. To identify eligible papers for thorough analysis, multi-phase searches and selections are accomplished. The study also includes extensive discussions on social media platforms and the technology, tools, and techniques used in analytics. The review discovered several methodologies and tools that were utilized with social media data. Descriptive analysis, regression analysis, and text analysis were the most commonly used analysis methods, while Facebook, Twitter, YouTube, Instagram, and Viber were the most popular social media networks. Current social media analytics research were noticed in a variety of domains, including marketing, education, politics, health, social, and business.
Impact of Service Quality Factors of Courier/Parcel Delivery Industry on Online Shopping Customer Satisfaction with Reference to SERVQUAL Model
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2023) Kodithuwakku, Supipi P. B.; Weerasekara, Dinusha S.
In the recent decade there has been a significant increase in e-commerce platforms within the Sri-Lankan context and with the outbreak of COVID- 19 the e-commerce businesses truly started to flourish and expand. E- businesses mainly use courier/parcel providers to engage in the last-mile delivery of the goods to the end customers, hence the courier services in a way act as an extension of the online brands. This study aims to identify which courier/parcel delivery service quality factors has a relationship between online shopping customer satisfaction in Colombo District with reference to the SERVQUAL model. With the reference of SERVQUAL model, the service quality factors that was relevant to the scope of the study was determined. Based on the review of the literature in this regard and with the use of convenience sampling technique, an online self- administered questionnaire was distributed among a sample of 250 within the Colombo District. The dimension empathy out of the four dimensions studied, appeared to have the highest correlation and regression, hence it is recommended that the courier/parcel delivery service providers prioritize it as a key factor when providing the courier services to the end customer. Further research is needed to identify the other service quality factors within the courier industry that could further strengthen the relationship with online shopping customer satisfaction by referring to more current literature.
Industry 4.0 Implementation in Sri Lankan Manufacturing Firms: A Lean Perspective
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2023) Bandara, Lahiru; Withanaarachchi, Amila; Peter, Suren
Manufacturing industries require the highest quality and efficiency throughout their value chain, to compete with countries having a labor cost advantage. Today, manufacturing firms are in a fast- phased run to automate their processes and increase value chain integration through advanced technologies. Industry 4.0 has gained traction within this community, where its components like IoT, Big data, and Cloud computing are being used by manufacturing firms to optimize and increase the efficiency of their workplaces. Obtaining the proper outcomes from these advanced technologies has been an issue for most of its users. Very few studies were found in the literature, that propose ways to mitigate the issues faced by these companies in their Industry 4.0 journey. Lean concepts are a popular and proven methodology used by firms worldwide to decrease the complexity and increase the productivity of their processes. Based on a systematic literature review, the study identifies the current knowledge on mitigating the barriers faced by manufacturing firms in Industry 4.0 implementations. To address the knowledge gap identified in the literature review, the study proposes and statistically tests a framework, on how the manufacturing environment can be improved to obtain the expected outcomes of Industry 4.0 implementations, through a lean theoretical lens. Thus, the stakeholders of the company can contribute towards successful implementations of Industry 4.0 while organizational processes are being standardized and optimized to integrate these advanced technological shifts.
Identifying the Factors for Influencing the Performance of the Virtual Teams in the Sri Lankan IT Industry
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2023) Samarakoon, W.S.M.S.S.; Dharmawansa, A.D.
The growth of virtual teams (VTs) in organizations can be attributed to the numerous technological advantages. However, virtual teams often face challenges when working on projects. Also, virtual teams are challenged not only to coordinate projects in virtual team environment but also to improve and build trust and, psychological safety within the culturally and geographically diverse team members. This research aims to identify the factors for influencing the performance of virtual teams in Sri Lanka, a topic that has received limited attention in developing countries. The study focuses on four independent variables; trust, knowledge sharing, psychological safety and team leader support with virtual team performance as the dependent variable. A 5 – point Likert-type online questionnaire was used to collect sample of 244 responses, representing 61 virtual teams from 22 private Information Technology (IT) companies in Sri Lanka. Structural Equation Modeling (SEM) was employed for data analysis. The findings indicate that trust and knowledge sharing significantly affect factors virtual team performance in the IT industry in Sri Lanka. However, knowledge sharing and team leader support were not found to have a significant impact on virtual team performance. Based on research findings, recommendations will be provided for IT industry employees and managers to ensure the trust and knowledge sharing between the team members, thereby improving overall team productivity.
Minimising Last-Mile Delivery Cost and Vehicle Usage through an Optimised Delivery Network Considering Customer-Preferred Time Windows
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2023) Abhilashani, G.Kasuri; Ranathunga, M.I.D.; Wijayanayake, A.N.
In the dynamic and developing e-commerce era, last-mile delivery has emerged as one of the critical operations among all. The last-mile delivery in the e-commerce industry is facing high costs due to a going economic crisis which led to fuel and other operating cost increments. To overcome this situation, the e-commerce industry needs to optimise vehicle delivery routing based on time windows to minimize the overall cost. Despite numerous studies on last-mile delivery, there is a paucity of studies on last-mile delivery optimization considering the customer's anticipated time windows. Therefore, this study has been conducted with the objective of optimizing and minimizing transportation costs and vehicle usage in last-mile delivery operations while meeting some practical requirements such as a variety of package types, package compatibility on different types of vehicles, customer expected delivery time windows, and a heterogeneous fleet of vehicles. After a careful literature review, this paper introduces a mathematical model to optimize last-mile delivery. The proposed mathematical model was simulated in SupplyChainGuru® modelling and simulation software. The study concluded that the overall last- mile delivery cost is minimized by about 22% while reducing the number of vehicles on the route, failed delivery package count and utilising the maximum possible capacity of vehicles while also increasing customer satisfaction by giving consumers a chance to select customer preferred time windows for package delivery. This cluster-based delivery will improve the routing of the e-commerce logistic supply chain and will serve as a platform for extending the cluster-based delivery process to other industries as well.
A Comprehensive Approach to Evaluating Software Code Quality Through a Flexible Quality Model
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2023) Shyamal, D.K.K.; Asanka, P.P.G.D.; Wickramaarachchi, D.
The rapid growth of the software engineering sector has led to a detrimental effect on the quality of software being developed. Code quality is crucial in determining the overall quality of software however, it is often observed that quality management programs primarily focus on internal processes within organizations, while the importance of code quality lacks proper attention despite the existence of quality standards for software products and processes. Due to its dynamic nature, the concept of quality poses a challenge in terms of precise definition, however, this paper addresses this issue by providing a comprehensive definition for code quality that considers all its dimensions, thus laying the foundation for conducting research related to quality. Code quality encompasses factors such as readability, scalability, performance, and adherence to industry standards. High-quality code is easy to understand, modify, and test, making it more reliable and less prone to bugs. By considering the multitude of challenges that currently exist and acknowledging the criticality of code quality, this study proposes an approach for assessing code quality, and a comprehensive quality model that considers the most critical code quality attributes and their relevant metrics along with corresponding threshold values specifically use in the contemporary software industry.
TQM Practices on Supply Chain Performance of Third- Party Logistics Services in Sri Lanka: The Moderating Role of Green Supply Chain Practices
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2023) Nawurunnage, K.; Prasadika, A.P.K.J.; Wijayanayake, A.N.
The growing need to address the threat of global warming and greenhouse gas emissions has placed immense pressure on logistics companies to adopt sustainable practices. With logistics operations being a significant source of greenhouse gas emissions, incorporating green supply chain management practices (GSCM) has become crucial to achieving environmental sustainability within the third-party logistics (3PL) industry. Exploring the existing literature under the concepts of Total Quality Management and Green Supply Chain Management reveals the need for future investigations into how those practices might potentially improve the logistics firm’s performance to achieve sustainability. Therefore, the main objective of this study is to identify the interrelationships of TQM practices and supply chain performance third- party logistics industry in terms of overall performance and identify the suitable TQM practices that can be applied to enhance the overall performance of Sri Lankan 3PLs and assess moderating effect of GSCM practices on that TQM- performance relationships. An online survey instrument was used to collect the data from executives, senior executives, and managers of 3PL firms in Sri Lanka. The statistical data analysis was done using PLS-SEM. The results found that top management support, customer focus, statistical process control, and continuous improvements are the significant total quality management practice for overall performance in the Sri Lankan 3PL industry. The study's findings are useful for the top management of 3PLs, policymakers, and academia to identify the level of GSCM implementation within the industry, and results provide insights into further considerations regarding the implementation of GSCM practices and TQM practices to achieve the supply chain performance of the 3PLs while achieving sustainability.
Determinants that Drive the Behavioural Intention of Employees in the IT industry to Use CI/CD Framework: A Study based on Sri Lankan IT Companies
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2023) Karunarathna, Chamindu; Jayasinghe, Shan; Wijayanayaka, W. M. J. I.
Continuous Integration and Continuous Delivery (CI/CD) is an Agile-based software development methodology becoming increasingly popular in the software development industry due to its ability to automate the software delivery process, reduce the time to market, and enhance software quality. However, despite the growing interest in CI/CD adoption, many organizations have not achieved full success in implementing and utilizing the CI/CD workflow. To address this gap, this study aimed to identify the factors that drive the behavioural intention of IT employees to use the CI/CD workflow: based on the Sri Lankan context. A systematic literature review using the PRISMA framework identified the Technology Acceptance Model and the Unified Theory of Acceptance and Use of Technology models as the most widely used and accepted models for understanding technology adoption. Therefore, TAM, UTAUT and past literature were used to develop the conceptual framework. The variables in this research model were measured through questionnaires with nominal and five- point Likert scales and close-ended questions, which were completed by the IT employees in Sri Lanka. Data cleaning and demographic data analysis were conducted using IBM SPSS 21, and preliminary data analysis was performed using PLS-SEM (SmartPLS 4). The study found that Performance expectancy is the most significant factor determining IT employees' behavioural intention to use CI/CD workflow. Therefore, the study concluded that organizations and management should focus more on enhancing employees' performance expectancy to adopt CI/CD workflow successfully.
Integrating Weather Patterns into Machine Learning Models for Improved Electricity Demand Forecasting in Sri Lanka
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2023) Abeywickrama, Shani; Asanka, P.P.G. Dinesh
The electricity demand in Sri Lanka is expected to increase steadily over time. Planning for future demand and ensuring an adequate electricity supply poses a significant challenge. It is crucial to accurately forecast the future demand in order to maintain an uninterrupted power supply. Previous studies have explored the correlation between weather factors and electricity demand with the aim of accurately predicting demand values. Thus, the objective of this study is to forecast the monthly electricity demand in Sri Lanka, by considering the influence of weather patterns. In this study, rainfall, humidity, and temperature weather parameters, along with historical monthly demand data, are taken into consideration. The identification of the most crucial weather variables is based on their correlation with electricity demand data. Various techniques have been employed for forecasting electricity demand over the past decade. However, the limitation of previous studies lies in their failure to incorporate past weather data alongside electricity demand data. This gap is addressed in the present study. This study used Vector Auto Regression (VAR) and Long Short-Term Memory (LSTM) models to forecast monthly electricity demand in each district of Sri Lanka. The VAR model demonstrated lower values by comparing the performance metrics, including Root Mean Square Error, Mean Square Error, Mean Absolute Error, and Mean Absolute Percentage Error. As a result, the VAR model was chosen as the most suitable model for forecasting monthly electricity demand by incorporating weather variables.
Defaulter Prediction in the Fixed-line Telecommunication Sector Using Machine Learning
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2023) Ginige, Sachini; Rajapakse, Chathura; Asanka, Dinesh; Mahanama, Thilini
In the modern connected era, the telecommunications sector plays a critical role in enabling efficient business operations across all industries. However, defaulting customers who fail to pay their dues after consuming services remain a significant challenge in the industry. Defaulters pose a risk to service providers, calling for measures to lessen both the probability of occurrence as well as its impact. Early identification of defaulters through prediction is a possible solution that enables proactive measures to mitigate the risk. However, the nature of the fixed-line product segment poses additional constraints in identifying defaulters, highlighting an existing knowledge gap. The research aims to evaluate the effectiveness of machine learning as a technique for the prediction of defaulters in the fixed-line telecommunication sector, and to develop an effective predictive model for the purpose. The success of machine learning techniques in analysis and prediction over traditional methods prompted its use in this study. The study followed the design science research methodology. An analysis was conducted based on past transaction data. Special consideration was given to the scenario of customers with little to no transaction history. Based on the analysis, a feature list for identifying defaulters was compiled, and multiple predictive models were developed and evaluated in comparison. The resulting predictive model, which uses the Random Forest technique, shows high performance in all considered aspects. The findings of the study demonstrate that machine learning techniques can effectively predict defaulters in the fixed-line telecommunication sector, with significant implications for mitigating the risk associated.
Forecasting of Medium-Term Energy Output of On-Grid Rooftop Photovoltaic Arrays -Case Study for a Sri Lankan Solar Panel Installer
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2023) Wickramasinghe, Bhagya; Asanka, P P G Dinesh
The world is shifting towards the higher utilization of renewable energy sources in the road to greener energy which conserves an environmentally friendly atmosphere. The generation of sustainable energy via adopting solar photovoltaic is common worldwide. The objectives of the research study are to identify the salient factors contributing to the energy generation of photovoltaic systems, to utilize a gamut of machine learning algorithms to build the predictive model and to identify the best machine learning algorithm to predict the energy generation based on accuracy and precision metrices. These objectives aid to achieve the aim of this study, which is to build a predictive model to determine the medium-term energy generated from on-grid rooftop solar systems. The study has unveiled a new piece of knowledge on how the photovoltaic system dynamics and location specific data has contributed to the prediction of the power output of the system. Further the findings are of paramount importance to the industry experts as well as the current and prospective solar panel users. The data of all solar panel sites of the installer was utilized and it was extracted from the source information systems. The necessary transformations and validations were applied and a detailed analysis was performed. The feature engineering, feature scaling, outlier-handling, multi-collinearity and feature selection was performed on data. The intended forecasting model based on fourteen supervised machine learning algorithms was built. The KNN Regression algorithm in the factor analysis of all features after principal component analysis has outperformed all other built models. Moreover, a strong positive co-relation was observed in the principal component analysis towards the solar panel energy output prediction. As part of future work, it’s imperative to build models utilizing a wider sample of on-grid roof top solar plants.
The Impact of Social Media Usage During Office Hours on Employee Performance: Evidence from a Sri Lankan Apparel Manufacturing Firm
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2023) Kothalawala, Charuka; Rathnayeka, Chaminda
Today, social media usage is an essential tool for communication among individuals and organizations. However, evidence suggests that some industry sectors are striving to understand the relationship between social media usage during office hours and job performance. In the Sri Lankan context, the apparel sector is struggling to understand this relationship. Thus, this study investigated the impact of social media usage on employee performance with special reference to a leading apparel manufacturing company in Sri Lanka. A deductive approach was adopted to conduct the research. Individual social media usage (ISM) and work-related social media usage (WSM) are considered as independent variables and employee job performance is the dependent variable. Findings suggest that ISM and WSM enhance the job performance of apparel industry workers in Sri Lanka. Furthermore, findings indicate that the apparel industry must not discourage social media usage during office hours, instead, must find methods of utilizing social media usage for the betterment of the firm. Practical and theoretical implications, limitations, and suggestions for future research are mentioned in the Discussion. Concluding remarks are discussed in the Conclusion.
An Automatic Density Cluster Generation Method to Identify the Amount of Tool Flank Wear via Tool Vibration
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2023) Adikaram, K.K.L.B.; Furukawa, Y.; Herwan, J.; Komoto4, H.
Determining the amount of tool flank wear (TFW) of a tool during operation is an important and cost-sensitive factor for maintaining the efficiency of the machine and product standards in Industry 4.0. Therefore, a variety of predictive analysis tools have been developed in this regard, with the objective of taking corrective action quickly and efficiently. In this paper, we present a TFW amount estimating method via plotting vibration generated during the cutting process on big data visualization and density cluster generation method known as Graphical Knowledge Unit (GKU). GKU generates density clusters by incrementing the RGB color values in the intersected markers due to data overlapping. In our previous work, the TFW amount of a cutting tool attached to a Computer Numerical Control (CNC) turning machine was checked. A workpiece of grey cast iron with an initial outer diameter of 110 mm was cut until it reached 60 mm. This process was repeated until the TFW amount, which was measured according to ISO 4288, met the recommended value range (0.3 ± 0.005 mm). After each cut, TFW amount and the surface roughness were measured following ISO 4288. Vibration was recorded using a triaxial accelerometer attached to the tool shank of the turning machine. In the present work, out of 29 cutting circles, vibration along the x-axis against vibration along the y-axis of selected cuttings were plotted using GKU. The density of the center of the plot (fixed point, FP) and the density of the highest density (dynamic point, DP) were measured using the color values of pixels as an index. The results showed a very strong linear correlation (0.95) between the TFW amount and vibration data density projected via pixel color values at FP. This shows that processing of vibration with GKU is a promising method to estimate TFW amount.
DrivEmo: A Novel Approach for EEG-Based Emotion Classification for Drivers
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2023) Gamage, T.A.; Sandamali, E.R.C.; Kalansooriya, Pradeep
Electroencephalogram (EEG) based emotion recognition approaches have proven to be successful with the latest technologies, and therefore, driver emotion recognition is also being widely discussed for enhancing road safety. This paper reveals a unique approach to driver emotion recognition for the calm, fear, sad, and anger emotional states where calm is the desired state of mind while driving. Emotiv EPOC X 14 channel EEG headset is utilised for the EEG collection, and ten subjects are involved in the experiment. EEG preprocessing of the collected EEG data is done using the EEGLAB toolbox in Matlab. EEG feature extraction is performed using Matlab, and feature selection and classification model training is done using the Classification Learner app in Matlab. ANOVA and ReliefF are employed as the feature selection algorithms, and Support Vector Machine (SVM) and Naïve Bayes classifiers are utilised for the emotion classification. The outcomes reveal that the highest mean accuracy of 95% is achieved from the Coarse Gaussian SVM classifier, while the lowest mean accuracy of 85% is obtained from the Fine Gaussian SVM classifier detecting the calm, fear, sad, and anger emotional states. In addition, all the other trained classifier models have an accuracy between 85% and 95%. Therefore, the findings suggest that the proposed EEG-based implementation approach of an emotion classification model for drivers is highly successful and can be employed in future research in the paradigm of driver emotion recognition as well. Besides, this research presents a critical literature review concerning critical aspects of EEG- based emotion recognition research.
Alapana Generation Using Finite State Machines and Generative Adversarial Networks
(Jayatharan Vithushigan; Alwis Dileeka (2023), Alapana Generation Using Finite State Machines and Generative Adversarial Networks, International Research Conference on Smart Computing and Systems Engineering (SCSE 2023), Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka. Page 6, 2023) Jayatharan, Vithushigan; Alwis, Dileeka
The recent advancements in deep learning techniques and computational power have promoted the development of novel approaches for music generation. In this study, generating alapana, an improvisational form of Carnatic music was proposed, by leveraging Generative Adversarial Networks (GANs) and Finite State Machines (FSM). The goal is to create melodious alapana sequences that follow a given input Raga, ensuring continuity and coherence throughout the generated musical piece. The proposed approach incorporates Carnatic music theory rules into the generation process to enhance the structural coherence of the generated alapana. Additionally, various hyperparameter settings were explored to achieve the best performance. The Fréchet Audio Distance, Percentage of Correct Pitches and the Subjective evaluation through human listeners are the evaluation metrics of this approach. The result of this study demonstrates the potential of using GANs and FSM for generating continuous and pleasing alapana sequences in Carnatic music, contributing to the growing body of research in computational music generation.
Effectiveness of Using Deep Learning for Blister Blight Identification in Sri Lankan Tea
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2023) Hewawitharana, G.H.A.U.; Nawarathne, U.M.L.A.; Hassan, A.S.F.; Wijerathna, L.M.; Sinniah, Ganga D; Vidhanaarachchi, Samitha P.; Wickramarathne, Jagath; Wijekoon, Janaka L.
Ceylon tea industry faces a major challenge in the form of pathogen-induced crop loss, with Blister Blight (BB) caused by Exobasidium vexans posing the greatest threat, leading to harvest losses of over 30%. This fungus attacks the tender tea shoots, resulting in a direct negative impact on the tea harvest. This paper presents a system to identify the suspicious tea leaves and BB disease at its early stages along with an assessment of severity, offering a potential solution to this critical issue. By utilizing real-time object detection, the system filters out non-tea leaves from the captured initial image of a segment of a tea plant. The identified tea leaves are then subjected to BB identification and severity assessment based on differing visual symptoms of the BB stages. This approach enables the system to accurately identify BB in the initial stage and severity stage, allowing for timely and targeted intervention to minimize crop losses. The YOLOv8 model has been able to correctly identify 98% of the objects it has detected as relevant (precision), and it has been able to correctly identify 96% of all the relevant objects present in the scene (recall). The Residual Network 50 (Resnet50) convolutional neural network (CNN) model was selected as the final model, achieving an accuracy of 89.90% during the training phase and an accuracy of 88.26% during the testing phase.
Sinhala Language Fake News Detection in Social Media Using Autoencoder-Based Method
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2023) Adihetti, Rahul; Jayalal, Shantha
The spread of fake news in the social media has grown significantly over the past few years. According to the New York Times, fake news is defined as "made-up articles meant to deceive." Additionally, the way they are released is almost identical to that of conventional news organizations. The issue is that a significant number of news outlets outside the major and reliable ones are disseminating unreliable information. This problem is exacerbated by the ease with which anything can be published from anywhere on well-known social networking and social media platforms. People can use this to their advantage by disseminating any type of message on various social networking sites to accomplish their objectives. In the Sri Lankan context, content posted in Sinhala greatly impacts fake news in Sri Lanka. Because utilizing the Sinhala language to describe emotions and feelings makes it easier to connect with Sinhala-speaking people than using content that has been published in other languages, like English. The use of Sinhala on social media has grown over the past few years. Additionally, as the use of the Sinhala language expanded, so did the number of occurrences of fake news. Based on the literature, approaches to identifying fake news depend on the features of the news content. Therefore, this research proposed an autoencoder- based method for Sinhala fake news detection, which is an unsupervised method. The method uses Text, User, Propagation, and Image features from the news content. And also, this research found the best feature combination to detect Sinhala language fake news content, which is a combination of Text, User, and Image features. The method gained an accuracy of 98% and 88% in Precision, Recall, and F1 Score by outperforming other existing anomaly detection methods. The main stakeholder of this study was fact-checking organizations in Sri Lanka.
Impact of Feature Selection Towards Short Text Classification
(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2023) Jayakody, J.R.K.C.; Vidanagama, V.G.T.N.; Perera, Indika; Herath, H.M.L.K.
Feature selection technique is used in text classification pipeline to reduce the number of redundant or irrelevant features. Moreover, feature selection algorithms help to decrease the overfitting, reduce training time, and improve the accuracy of the build models. Similarly, feature reduction techniques based on frequencies support eliminating unwanted features. Most of the existing work related to feature selection was based on general text and the behavior of feature selection was not evaluated properly with short text type dataset. Therefore, this research was conducted to investigate how performance varied with selected features from feature selection algorithms with short text type datasets. Three publicly available datasets were selected for the experiment. Chi square, info gain and f measure were examined as those algorithms were identified as the best algorithms to select features for text classification. Moreover, we examined the impact of those algorithms when selecting different types of features such as 1-gram and 2-gram. Finally, we look at the impact of frequency-based feature reduction techniques with the selected dataset. Our results showed that info gain algorithm outperform other two algorithms. Moreover, selection of best 20% feature set with info gain algorithm provide the same performance level as with the entire feature set. Further we observed the higher number of dimensions was due to bigrams and the impact of n grams towards feature selection algorithms. Moreover, it is worth noting that removing the features which occur twice in a document would be ideal before moving to apply feature selection techniques with different algorithms.