Authors - Valencia Vannessa Taslim, Melissa Anastasia, Shalva Andena Rizaldi, Tiurida Lily Anita Abstract - Online hospitality reviews provide valuable insights into guest experiences, service quality, and operational performance. However, the unstructured and noisy nature of review text makes large-scale analysis difficult, especially for Indonesian-language reviews that often contain informal expressions, abbreviations, spelling variations, and inconsistent sentence structures. Although sentiment analysis has been widely applied in hospitality research, studies focusing on Indonesian-language hospitality reviews remain limited, and few have presented a reproducible Natural Language Processing (NLP) workflow for multiclass sentiment classification. This study proposes a reproducible Indonesian NLP pipeline for classifying hospitality reviews into positive, neutral, and negative sentiment categories. The workflow integrates review collection, sentiment annotation, Indonesian text preprocessing, TF-IDF feature extraction, and super-vised classification using Naïve Bayes, Support Vector Machine (SVM), and Logistic Regression. The dataset consists of 450 Indonesian-language hotel reviews collected from Google Reviews across three hotel segments: budget, mid-scale, and upscale. The experimental results show that SVM achieved the best overall performance, with 91.78% accuracy, 91.35% precision, 91.78% recall, and 91.50% F1-score, outperforming Naïve Bayes and Logistic Regression under the same experimental setting. These findings indicate that classical machine learning, when supported by systematic preprocessing and consistent feature representation, remains highly effective for Indonesian hospitality review analytics. This study contributes a practical and reproducible baseline for Indonesian-language sentiment classification and provides a foundation for future intelligent review monitoring systems in the hospitality sector.
Authors - Jolou Vincent M. Jala, Everly A. Nacalaban, Nenon Roy A. Sandinao, Erlinda D. Rivera, Hilfiger L. Cubarrubia Abstract - This article explored the adoption of Artificial Intelligence in Financial Management Systems of Higher Education Institutions (HEIs) by utilizing a systematic review of related literature. The study focuses on reviewing pre-sent literature on Artificial Intelligence adoption in financial management systems, recognizing the benefits of AI integration, scrutinizing the challenges and barriers to implementation, and offer recommendations for effective and successful AI integration in HEIs. The findings disclosed that artificial intelligence has the capability to meaningfully enhance financial management systems in Higher Education Institutions through automated financial reporting systems, budgeting forecasting and predictive analytics, fraud detection and risk management, and expense tracking and optimization. Adoption of Artificial Intelligence improves efficiency, enhances accuracy, provides better decision-making and cost optimization. More-over, it enhances operational efficiency by systematizing monotonous financial tasks, enhances accuracy by plummeting human faults, supports better decision-making through actual financial data and predictive analytics, and helps to long-term cost optimization and financial sustainability. These improvements permit institutions to alter from manual and volatile financial management routines toward more data-driven, calculated and strategic, financial planning and re-source provision. Conversely, the study also found several challenges that deter AI adoption in Higher Education Institutions, specifically in developing countries such as the Philippines. These challenges include high initial investment and maintenance costs, limited technical skills among staff, data privacy and cybersecurity risks and resistance to organizational change. Numerous HEIs are still in the developing stage of digital transformation and depend chiefly on enterprise systems and basic accounting rather than advanced Artificial Intelligence technologies. The article concludes that successful AI in- corporation requires institutional readiness, strategic planning, capability building, infrastructure progress, and robust data governance policies to completely maximize the advantages of Artificial Intelligence in financial management systems.
Authors - Denver Novencido Abstract - An organization’s operational efficiency, productivity, and reliability can be adversely affected by using manual-based systems. Some of the issues associated with using a manual-based approach include inefficient processes, inconsistent documentation, difficulty in monitoring and validating records, and limited accessibility. The development of information systems provides a solution to address the limitations and challenges of a manual-based approach in organizations. This study presents the design and implementation of a cloud-based information system integrated with decision support capabilities to streamline organizational operations, enhance data storage and retrieval, and facilitate strategic planning. The system was created using the Agile Unified Process (AUP) software development methodology. Evaluation results indicate strong compliance with ISO software quality standards, making it a suitable tool for managing organizational operations.
Authors - Felix Kabwe, Jackson Phiri Abstract - This study explores how blockchain-based Identity and Access Management (IAM) systems can enhance the security and efficiency of Digital Financial Services (DFS). As DFS environments grow more complex and involve multiple stakeholders, traditional IAM systems face challenges such as centralization, limited interoperability, and scalability constraints. Blockchain offers a compelling alternative by enabling decentralized, transparent, and tamper-resistant identity management. The study compares three main IAM models: centralized systems supported by blockchain, federated identity management, and Self-Sovereign Identity (SSI). Using the Technology-Organization-Environment (TOE) framework alongside a semi-quantitative scoring approach, the research evaluates these models across key factors including security, privacy, usability, scalability, governance, cost, and regulatory alignment. The findings highlight clear trade-offs. Centralized systems excel in performance, cost efficiency, and regulatory compliance but are vulnerable to single points of failure. Federated models strike a balance by improving interoperability and user experience, though they introduce governance complexity. SSI provides strong privacy and user control but faces challenges in usability, scalability, and regulatory acceptance. Overall, no single model fully meets DFS needs. Federated systems are currently the most practical, while hybrid federated–SSI approaches offer the most flexible, scalable, and user-focused solution.
Authors - Md. Monowar Hossain, Fahima Hossain, Md. Shahidul Islam, Md. Tanvir Ahmed, Reduan Ahmed Abstract - This automated image captioning is on one hand a Computer Vision (CV) and Natural Language Processing (NLP) application, but on the other hand, conventional CNN-RNN models suffer from feature loss and long-range dependency. The proposed model in this study is a parameter balanced multi-modal model that consists of a dual-encoder network which combines Effi-cientNet-B4 for hierarchical features and MobileNetV2 for geometric efficiency, as well as a multi-head Transformer decoder. The model was evaluated on Flickr8k, and tested with a dynamic scalar weight mechanism and teacher-forced optimization, the BLEU-1 was 0.5774 and METEOR was 0.4129. Interestingly, the ablation results also showed that although the dual-encoder method is competitive, the pathway of the standalone MobileNetV2 is slightly better than the fused pathway in terms of BLEU-4 (0.2284 vs. 0.20). This indicates that the pathway may be redundant during the concatenation process. This study validates the possibility of using Transformer decoders instead of RNN bottle-necks and offers important considerations for the optimization of real-time feature fusion for vision tasks.
Authors - Aarthi R, Aniketha Prasad, Dhamini Manoj, Manasvi G, Meghaa Sunil Abstract - Early and accurate diagnosis of dermatological disorders remains one of the main issues in clinical dermatology, especially with regard to diseases with similar appearances. Despite the achievements of deep learning methodologies in the classification of cutaneous lesions with the help of images, structured clinical metadata is not used to the fullest, despite its significant diagnostic potential. In a practical clinical setting, dermatologists do not solely use visual evaluation of the case but also use patient-specific metadata, which includes age, lesion progression, pruritus, hemorrhage, anatomic location, prior biopsy, and family history. The current study presents a fully explainable, metadata based, multi-class classification of skin diseases, using the PAD-UFES-20 database, and concentrated on 6 distinct diagnostic categories. Although the dataset is dermoscopic, the predictive quality of formal metadata variables are mainly under consideration in the present work. The explainability analyses revealed that biopsy status, elevation, itch, region and age are attributes that have significant effects on the classification results. However, empirical evidence shows that the reduced model consisting of the premier five features lowers accuracy, which highlights the importance of a thorough combination of metadata features to determine skin disease rather than limited combination. Comparative studies indicate that the Multi-Layer Perceptron shows an improvement in a model performance with a corresponding increase of the number of selected features. The suggested framework thus highlights interpretability in line with predictive efficacy thus enhancing the importance of transparent artificial intelligence systems in medical decision-making.
Authors - Veeravalli Sri Satya, Anjan Babu G Abstract - Human-computer interaction with smart consumer electronics predominantly requires physical peripherals, which introduce limitations regarding hardware degradation, shared-surface hygiene, and usability in hands-free environments. Voice-activated systems provide an alternative but exhibit high latency and degraded performance under ambient noise. This paper presents a multi-layered touchless gesture control framework that translates human hand kinematics into direct system actuation. The architecture utilizes a standard web camera and the Google MediaPipe framework to extract 21 three-dimensional hand landmarks in real time. To bypass the computational bottlenecks of traditional Convolutional Neural Networks (CNNs), the system employs a custom heuristic algorithm to classify eight distinct static and dynamic gestures by analyzing the geometric relationships between finger joints. The framework processes these classifications locally and actuates Android-based Smart TVs over Wi-Fi utilizing Android Debug Bridge (ADB) protocols [11]. Evaluated in a controlled environment, the pipeline achieved an average processing time of 35 milliseconds per frame (approximately 30 frames per second) with a network transmission delay of 50 to 80 milliseconds. The results suggest that computationally lightweight computer vision models, when paired with structured state-machine logic, can effectively replace physical remote controls without requiring dedicated GPU hardware.
