Recent Submissions
BookStore: An Innovative Book Recommendation System Driven by Facial Expressions
(Department of Industrial Management, Faculty of Science, University of Kelaniya., 2025) Neelananda, M.; Rathnayake, R.
In the age of digital content consumption, personalization plays a crucial role in enhancing the user experience. However, many existing recommendation systems focus primarily on past interactions or ratings, often overlooking the user's current emotions and cultural background. This is particularly relevant in literature, where readers' mood and background can significantly influence their reception of various genres or topics. The proposed study addresses this gap by proposing an emotion-based book recommendation system that utilizes Facial Emotion Recognition (FER) to suggest culturally appropriate books aligned with the user’s emotional state. A deep learning model based on Convolutional Neural Networks (CNN) was trained to classify facial expressions and identify a person’s real-time emotions. The system features a culturally diverse dataset of books, categorized by emotion-relevant themes using Natural Language Processing (NLP) techniques. The user’s emotional state is determined through the FER model, and recommendations are made based on this identified mood and the user's cultural background. The model is trained to recognize facial emotions using labeled facial emotion datasets and has been tested across multiple scenarios to ensure it provides relevant book suggestions. The prototype system accurately identified users’ emotions with an 84% success rate and successfully recommended books of interest in 90% of the attempts. Users confirmed that the system could enhance their interest in reading by aligning book suggestions with their emotional and cultural inclinations. While these preliminary results are promising, they highlight the model’s potential to create a friendly reading environment by suggesting books that resonate with users' moods and cultural contexts.
Predictive Modeling of Rubber Plant Growth Using Environmental Data and Machine Learning Techniques
(Department of Industrial Management, Faculty of Science, University of Kelaniya., 2025) Liyanage, Y. L. S. N.; Hettiarachchi, P. L.; Dilpashan, L. T. C.; Navodya, M. L. H.; Jayasekara, B.
Rubber plant nurseries require effective management to maximize agricultural resources and ensure good plant health. This paper presents the Rubber Plant Growth Prediction System, which uses environmental data to predict the diameter of the stem, an important index used in plant growth. In this work, initial plant growth was predicted using a vision-based technique, but due to difficulties in data acquisition and the accuracy point of view, that approach may not be feasible. The method based on stem diameter was then adopted, incorporating environmental parameters such as soil moisture, soil temperature, ambient temperature, and humidity. With a limited dataset, it was quite challenging to be accurate in the predictions, but developing a comprehensive data collection system filled those gaps, making the measurements more reliable. This forms the basis for increasing predictive accuracy in improving resource management in the nursery and therefore contributing to sustainable agricultural practices.
Determining Factors Related to Artificial Intelligence Adoption Among Sri Lankan ICT Service Providers
(Department of Industrial Management, Faculty of Science, University of Kelaniya., 2025) Nizar, Z.; Lakshan, J.; Wijayarathne, C.; Jayasuriya, N.; Oshani, W.; Rathnapriya, S.
This study investigates the factors influencing Artificial Intelligence (AI) adoption among ICT service providers in Sri Lanka, a developing economy where AI integration remains in its early stages. Using the Technology-Organization-Environment (TOE) framework, six key determinants were examined: relative advantage, data quality, top management commitment, employee adaptability, competitive pressure, and external support. Data was collected via a structured questionnaire distributed to 146 ICT service providers, with the sample size determined using G*Power software for statistical robustness. Ordered probit regression was employed to analyze the data, providing precise insights for ordinal variables. The findings identify data quality (DQ) and relative advantage (RA) as significant drivers of AI adoption, underscoring the critical role of high-quality data and the operational benefits of AI technologies. However, top management commitment (TMC) exhibited a negative impact, highlighting barriers in leadership awareness and alignment with AI strategies. Although competitive pressure (CP), external support (ES), and employee adaptability (EA) were not statistically significant, they showed potential as mediators or moderators in specific contexts. This study bridges a critical gap by providing localized insights into AI adoption challenges and opportunities in Sri Lanka. It emphasizes the importance of data management, leadership commitment, and strategic alignment, offering actionable recommendations for policymakers and industry leaders to enhance competitiveness and digital transformation in the global economy.
Advanced NLP Framework for Analysing Elasticity Feedback in Apparel Design
(Department of Industrial Management, Faculty of Science, University of Kelaniya., 2025) Wimalasuriya, D. T.; Rajapakse, C.; Jayatissa, Y.
Elasticity is critical in modern apparel design, directly affecting product performance and customer satisfaction. This study presents a scalable framework leveraging advanced Natural Language Processing (NLP) to systematically analyze elasticity-related customer feedback. Utilizing 93,860 customer reviews of leggings, pants, and crops, the framework combines manual annotation, keyword extraction, and Named Entity Recognition (NER) to identify and refine elasticity-specific reviews. A GPT-4o–based categorization model and hierarchical clustering of themes further segment reviews into aspects: attributes (e.g., flexibility, durability), components (e.g., stretch fabric, elastic waistbands), usage situations (e.g., washing, wearing), and common issues (e.g., pilling, sagging). Aspect-Based Sentiment Analysis (ABSA) and VADER scoring capture explicit and implicit sentiments, while advanced visualizations (heatmaps, scatter plots) reveal correlations among product features, consumer sentiment, and usage contexts. Comparative evaluations demonstrate the proposed framework’s superiority over lexicon-based and other machine learning methods. The results, integrated with product metadata, provide actionable insights for manufacturers, highlighting the importance of activity-specific designs and innovative fabric development. This work establishes a foundation for future integration of multimodal data, further enhancing customer feedback analytics in the apparel industry.
Deep Learning-Based Approach for Distinguishing Between AI-Generated and Human-Drawn Paintings
(Department of Industrial Management, Faculty of Science, University of Kelaniya., 2025) Warnakulasooriya, A. I.; Rupasingha, R. A. H. M.; Kumara, B. T. G. S.
With the increasing number of robust Artificial Intelligence (AI) art generation applications, more realistic AI-generated paintings (AIGPs) are emerging, creating a significant impact on artists. Due to the widespread acceptance of AIGPs, the cultural, historical, and monetary value of real masterpieces is becoming uncertain, raising concerns about the significance of human painters and their artistic techniques. To protect artists’ rights, it is crucial to differentiate AIGPs from human-drawn paintings (HDPs). Accordingly, the main objective of this research is to develop a Convolutional Neural Network (CNN) model that can automatically distinguish between AI-generated and human-drawn paintings without human intervention. Unlike previous studies that focused mainly on pixel-level analysis, the proposed model considers additional features such as edge patterns, object arrangements, pattern distributions, and gradient characteristics in painting classification. A diverse dataset of 3,000 paintings from the AI-ArtBench Dataset—comprising 1,500 AIGPs and 1,500 HDPs across 10 different art themes—was collected and preprocessed for this study. The AIGPs were generated in equal proportions using Latent Diffusion and Standard Diffusion Models. The implemented CNN model achieved an optimum classification accuracy of 90% with a training data size of 10%, while the ANN model exhibited 77% accuracy under the same conditions. Furthermore, models were compared using performance metrics such as precision, recall, F1-score, RMSE, and MAE. Through Gradient-weighted Class Activation Mapping (Grad-CAM), the key visual features that the CNN model used to distinguish AIGPs from HDPs were identified. These findings highlight the potential of automated systems in detecting AI-generated versus human-created artworks for authentication purposes. Future work will focus on analyzing model performance across different art styles and identifying the unique discriminative features associated with each.