Software Engineering Teaching Unit

Permanent URI for this collectionhttp://repository.kln.ac.lk/handle/123456789/26470

Browse

Search Results

Now showing 1 - 10 of 23
  • Item
    Leveraging Artificial Intelligence for Ethical Social Media Influencer Communication
    (2024) Hewapathirana, I. U.
    This chapter explores the connections between artificial intelligence (AI) and the ethical dimensions of influencer communication on social media. The ethical aspects are evaluated according to the criteria outlined in the Professional Code of Ethics of the Public Relations Society of America (PRSA). The study reviews the multiple aspects of influencer communication, including emerging challenges and legal implications resulting from the continued development of AI in social media. Furthermore, a dataset was collected from the social media platform Reddit, and a case study analysis was performed using the NodeXL software. This empirical investigation aims to investigate social media users' perspectives on specific ethical concerns associated with integrating artificial intelligence (AI). The findings presented in this chapter provide scholars with an advanced understanding of AI capabilities, offer industry professionals valuable guidance for ethical decision-making, and offer lawmakers guidance for developing regulatory frameworks.
  • Item
    TourismXplorer: Interactive Dashboard for Data-Driven Decision-Making in Sri Lanka’s Tourism Industry
    (2024) Thilakarathna, W. A. S. M. S.; Hewapathirana, I. U.
    Abstract: The tourism industry is a critical component of Sri Lanka’s economy, necessitating advanced tools for data-driven decision-making to enhance strategic planning and operational efficiency. This study presents the development of a comprehensive tourism dashboard designed specifically for tourism businesses in Sri Lanka. The dashboard offers a holistic view of the tourism landscape by integrating diverse data sources, including annual statistical reports (2018-2023), climate variables from the Sri Lanka Meteorological Department, and TripAdvisor reviews. The novelty of this research lies in its multifaceted data integration, advanced visualization techniques, and predictive analytics capabilities. The dashboard provides stakeholders with real-time and historical insights into tourism dynamics. It includes key performance indicators (KPIs) such as tourist arrivals, revenue, expenditure, accommodation statistics, climate impact, visitor demographics, and sentiment analysis from reviews. Visualizations range from line, pie, and bar charts to shape maps, heat maps, and word clouds, enhancing data accessibility and interpretability. A standout feature of the dashboard is its predictive analytics page, which allows users to forecast tourist arrivals based on selected explanatory variables such as climate data and customer sentiments. This predictive ability enables stakeholders to simulate various scenarios and better prepare for future trends, making the dashboard an invaluable tool for strategic decision-making. The dashboard’s user-friendly interface and customizable filtering options allow users to tailor their analyses based on specific criteria, such as year, region, and visitor attributes. This targeted approach ensures that tourism businesses can leverage the dashboard for practical decision-making, aligning with sustainable tourism development goals by monitoring environmental and social impacts. This research advances the field of tourism analytics and provides a practical tool for enhancing the strategic and operational capabilities of tourism businesses in Sri Lanka. Future enhancements may include the incorporation of more sophisticated predictive models, which would further improve the dashboard’s utility.
  • Item
    Development of a machine learning model for air quality forecasting: leveraging long-term meteorological data analysis to predict air quality index in Colombo District
    (2024) Rathnayaka, R.M.S.I.; Hewapathirana, I. U.
    Air quality is a critical aspect of environmental health, directly impacting individuals and the broader ecosystem. Therefore, real-time monitoring and understanding the factors influencing air quality are crucial. The most typical reasons for air pollution are vehicle emissions, organic waste burning, and petroleum refining. However, other factors have arisen as causes of air pollution. Although meteorological factors are natural phenomena, they have been changing detrimentally due to human actions. Extreme meteorological events may significantly influence air quality. In Sri Lanka, a region with its own set of environmental challenges, understanding the dynamics of air quality is important. Over the past decade, Sri Lanka has witnessed notable shifts in weather patterns, with potential implications for human well-being. Available data indicates that Colombo often experiences high levels of air pollution. Recognizing these factors, this research introduced a model for real-time forecasting of the Air Quality Index (AQI) based on meteorological factors, emphasizing the Colombo district. The research focused on the period from 2020 to 2023, using a dataset that includes daily meteorological factors, wind speed, temperature, atmospheric pressure, rainfall, and relative humidity, alongside daily AQI values for the Colombo district. A temporal analysis identifies long-term trends and patterns in air quality. The study leveraged five machine learning algorithms: Linear Regression, Random Forest Regression, Gradient Boosting Regression, Support Vector Regression, and Long Short-Term Memory Network to develop models for predicting air quality based on meteorological factors. It also evaluated the performance of these machine learning models using metrics such as Mean Squared Error, Root Mean Squared Error, Mean Absolute Error, and R-Squared to determine each model’s reliability in predicting the AQI. In conclusion, this research aims to discuss the role of weather variables in shaping air quality in the Colombo district. The outcomes contribute to understanding air quality in Sri Lanka and the broader global discourse on utilizing advanced technologies for environmental monitoring and forecasting. With insights into the predominant weather factors influencing air quality, decision-makers can formulate policies to improve the region’s air quality based on seasonal weather pattern changes.
  • Item
    Deep learning-based correctness assessment for the Tadasana (Mountain Pose) Yoga Asana
    (2024) Gayan, V.G.N.; Hewapathirana, I. U.
    Yoga has become increasingly popular worldwide, but practicing without proper guidance can lead to incorrect posture alignment, reducing effectiveness, and increasing injury risk. This research aimed to address this issue by developing a deep learning-based system that relies on the MediaPipe framework to assess the correctness of the Tadasana yoga asana and provide real-time feedback for improvement. A deep learning-based system was selected for the proposed study to implement the MediaPipe framework, for its outstanding real-time performance (75.9% mean average precision on the COCO dataset) and cross-platform efficiency. Using MediaPipe, a custom-developed web app analyzed more than 50 professional yoga instruction videos to extract crucial body angles for each Tadasana step, generating the dataset for the yoga pose angle calculation algorithm. This approach accounts for MediaPipe’s inherent variability in landmark detection, ensuring robust angle calculations. The primary goals of this study were to develop an accurate pose estimation and angle calculation algorithm specifically optimized for Tadasana, as well as a comprehensive, real-time feedback mechanism for pose correction. The proposed system integrated MediaPipe’s pose estimation capabilities with a custom angle calculation algorithm and a rule-based feedback system. An extensive evaluation was conducted using more than 100 images of correct and incorrect poses for each of the three Tadasana steps. The system demonstrated promising results, achieving accuracy scores of 78, 75, and 72% for steps 1, 2, and 3, respectively. It was observed that the system’s performance varied based on factors such as image quality and environmental conditions. This study demonstrates the feasibility and potential of using deep learning and computer vision techniques for precise yoga pose correction. Future work will focus on enhancing the system’s robustness across diverse conditions, expanding its capabilities to encompass a wider range of yoga poses, and implementing real-time video analysis for feedback generation. These advancements could significantly enhance the accessibility and effectiveness of remote yoga instruction, making proper technique more attainable for practitioners.
  • Item
    Empowering influence discovery: Utilizing machine learning for social media influencer identification
    (2024) Devyanjalee, D.D.W.N.; Hewapathirana, I. U.
    In today’s dynamic digital landscape, influencer marketing has become a cornerstone of marketing strategies, leveraging social media platforms to engage with audiences. Accurately identifying influencers within social media platforms poses a formidable challenge. Traditional machine learning approaches relying solely on metrics such as network analysis and user profile data, often fall short in capturing the dynamics of influencer impact and resonance with audiences. To address this gap, this study aimed to enhance influencer identification accuracy by leveraging both user profile and engagement metrics alongside text analysis. The methodology adopts a sequential explanatory design, combining quantitative analysis of user profile metrics with qualitative analysis of text-related factors. Data collection from social media platforms, particularly X, comprises user profile and social data. The quantitative phase employs established algorithms like the PageRank algorithm to identify top influencers based on user profile data, while machine learning models, logistic regression, decision trees, and random forest are trained using user profile data to discern influential user profiles. The qualitative phase involves text analysis techniques, including keyword matching and lemmatization, to extract valuable insights from tweets. Machine learning models are then trained using both user profile and social data alongside text analysis data to discern influential user profiles. The models are then compared to assess the impact of incorporating engagement metrics with text analysis. Findings from this study indicate that while user profile metrics alone exhibit high accuracy in influencer identification, with the random forest model achieving an F1 score of 0.90, the incorporation of engagement metrics introduce complexities affecting model performance, resulting in an F1 score of 0.70. The random forest model emerges as the most robust performer, maintaining high accuracy despite these challenges. This research contributes to advancing influencer identification strategies within digital marketing, offering insights into the effectiveness of integrating both user profile and engagement metrics with text analysis for capturing the true essence of influencer influence and resonance with audiences. The findings underscore the challenges of leveraging engagement metrics for influencer identification and highlight the need for further refinement of methodologies to empower marketers in navigating the complexities of the ever-evolving digital landscape.
  • Item
    A Case Study in Financial Fraud Detection using Big Data Analytics
    (2021) Boteju, W. P. A.; Hewapathirana, I. U.
    The financial industry is currently undergoing digital transformations across products, services and business models. This digitization is aimed at automating most of the manual financial transactions and other relevant services. Therefore, spotting fraud in financial transactions has become an important priority for all financial institutes. With the advances in modern technology and global communication, fraud has increased significantly, causing great damages. The focus of this paper is to experiment different approaches for detecting fraudulent activities in a real-world dataset of financial payment transactions. The dataset is obtained from Kaggle and consists of 6 million transaction records and 10 features with the transaction label as ‘fraudulent’ or ‘non-fraudulent’. These features are investigated using exploratory data analysis and only 6 are retained for the experiment such as payment-type, account-balance, transaction-amount etc. Two supervised machine learning algorithms, the random forest and the support vector classifier are employed for detecting fraudulent transactions. The dataset is large and requires high computational power to process and train machine learning algorithms. Furthermore, another challenge is the highly imbalanced distribution between fraudulent (0.1%) and the non-fraudulent (99.9%) classes. The goal of this research is to solve both these issues. In order to handle class imbalance, the effect of oversampling the minority class data using the synthetic minority oversampling technique (SMOTE), and undersampling the majority class using random undersampling are investigated. Computational efficiency is achieved through the Apache Spark implementation, which provides distributed processing for big data workloads. The best performance is obtained using the random forest algorithm on the oversampled dataset with an accuracy of 99.95%, F1-score of 0.9994, recall of 0.9994, Geometric mean of 99.94% and a model training time of 13.9 minutes. This paper provides valuable insights on dealing with large scaled highly imbalanced big datasets for predicting financial frauds and generating alerts.
  • Item
    A Systematic Investigation on the Effectiveness of the Tabbert Model for Credit Card Fraud Detection
    (2022) Hewapathirana, Isuru Udayangani; Kekayan, Nanthakumar; Diyasena, Deshanjali
    As a result of rapid digitisation, online transactions using credit cards have become popular. With this, fraudulent activities have also increased considerably. Although many supervised and unsupervised machine learning techniques were proposed in past research for identifying fraudulent transactions, they do not fully utilize the tabular and hierarchical structure present in transaction datasets. Recently, the TabBERT neural network model was proposed to calculate row-wise embeddings that capture both inter and intra dependencies between transactions in tabular time series data. In this research, we present a systematic experimental framework to assess the effectiveness of applying the embeddings calculated using the TabBERT model for credit card fraud detection. We employ the calculated row embeddings for fraud detection using three unsupervised machine learning algorithms and two supervised machine learning algorithms. We perform our experiments on a synthetic dataset that has been generated using the TabGPT model. Overall, TabBERT-based embeddings increase the performance of the supervised learning models with the extreme gradient boosting model achieving a precision of 99% and an F1 score of 98%, and the multilayer neural network model achieving a precision of 97% and an F1 score of 95%. For unsupervised learning, the use of TabBERT embeddings increases the recall rate of K-means clustering algorithm by 0.19%.
  • Item
    A Review of Recent Trends in Sri Lankan Social Media Analytics Research
    (2023) Hewapathirana, Isuru Udayangani; Sandaruwani, M. D.
    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.
  • Item
    Navigating the age of AI influence: A systematic literature review of trust, engagement, efficacy and ethical concerns of virtual influencers in social media
    (Hewapathirana IU, Perera N. (2024). Navigating the age of AI influence: A systematic literature review of trust, engagement, efficacy and ethical concerns of virtual influencers in social media. Journal of Infrastructure, Policy and Development. 8(8): 6352. https://doi.org/10.24294/jipd.v8i8.6352, 2024) Hewapathirana, Isuru Udayangani; Perera, Nipuni
    This systematic literature review (SLR) delves into the realm of Artificial Intelligence (AI)-powered virtual influencers (VIs) in social media, examining trust factors, engagement strategies, VI efficacy compared to human influencers, ethical considerations, and future trends. Analyzing 60 academic articles from 2012 to 2024, drawn from reputable databases, the study applies specific inclusion and exclusion criteria. Both automated and manual searches ensure a comprehensive review. Findings reveal a surge in VI research post- 2012, primarily in journals, with quantitative methods prevailing. Geographically, research focuses on Europe, Asia Pacific, and North America, indicating gaps in representation from other regions. Key themes highlight trust and engagement’s critical role in VI marketing, navigating the balance between consistency and authenticity. Challenges persist regarding artificiality and accountability, managed through brand alignment and transparent communication. VIs offers advantages, including control and cost efficiencies, yet grapple with authenticity issues, addressed through human-like features. Ethically, VI emergence demands stringent guidelines and industry cooperation to safeguard consumer well-being. Looking ahead, VIs promises transformative storytelling, necessitating vigilance in ethical considerations. This study advocates for continued scholarly inquiry and industry reflection to navigate VI marketing evolution responsibly, shaping the future influencer marketing landscape.
  • Item
    Modeling and Forecasting Mortality in Sri Lanka
    (2014) Aberathna, Wasana; Alles, Lakshman; Wickremasinghe, W. N.; Hewapathirana, Isuru
    The purpose of this study is to develop sex-specific mortality estimation models using historical mortality data for Sri Lanka, based on the statistical time series techniques attributed to Lee and Carter (1992). Historical mortality data was analyzed in the light of significant historical episodes. Several alternative univariate time series models were examined for modeling males and females, as well as a bivariate vector autoregressive (VAR) model. The VAR model when fitted to the first differenced series performed better than the univariate models and hence used for forecasting purposes. From the estimated VAR model, mortality forecasts were generated for the period up to 2030 and life tables were generated for the selected periods of 2006-2008.