1. Non-Autoregressive and Iterative Refinement Non-Autoregressive Transformer with additional Graph layer for Urdu-to-English Machine Translation

Authors and affiliations

• Huma Israr — Department of Computing and Emerging Technology, National Skills University Islamabad, Pakistan

• Shahid Anwar — Department of Computing and Emerging Technology, National Skills University Islamabad, Pakistan

• Mohamad Fadli Zolkipli — School of Computing, Universiti Utara Malaysia, Malaysia

• Arslan Ahmed — Sheffield Hallam University, United Kingdom

Abstract

Neural Machine Translation (NMT) has achieved state-of-the-art performance using autoregressive Transformer. However, the trade-off between translation quality and decoding speed remains a central challenge. This problem is further intensified for low-resource languages such as Urdu, where limited parallel data and complex linguistic structure result in poor translation quality and reduced model effectiveness. To overcome this problem Non-Autoregressive Transformer (NAT) and Non-Autoregressive Transformer with Iterative Refinement (iNAT) are proposed. This paper investigates NAT and iNAT for low resource language Urdu-English machine translation. We analyze translation accuracy and their effectiveness for low-resource language pairs. Experimental results show that both NAT and iNAT outperform the baseline Autoregressive Transformer (ART) in terms of translation accuracy. However, the overall translation quality remains limited due to the complex linguistic structure and resource scarcity of the Urdu-English pair. To address this limitation, we further refine these architectures by introducing an additional graph-based learning layer that captures latent relationships among tokens through attention-driven graph representations. The proposed graph-enhanced model learns contextual dependencies by dynamically constructing graphs over attention representations. Experimental evaluation demonstrates that the proposed architecture improves the translation performance of both NAT and iNAT models. On the Urdu-English translation task, the proposed model achieves a 3.0% higher BLEU score than the baseline models. The findings highlight the significance of incorporating graph-based layer to enhance translation quality, making NMT systems more effective for low-resource and linguistically complex languages such as Urdu.

Keywords: Low-Resource language, Machine Translation, Non-autoregressive, Transformer, URDU.

2. Secure Multi-Disease Detection Using Federated Learning with Optimized Graph-Based Deep Feature Attention with Blockchain Support

Authors and affiliations

• Krishna Prasad N Rao — School of Computer Science & Engineering, Reva University, India

• Selvan Chinnaiyan — School of Computer Science & Engineering, Reva University, India

Abstract

The rise of digital healthcare systems has generated large, fragmented medical data for enhanced illness prediction. However, modern centralized machine learning algorithms distribute sensitive patient data, raising privacy, security, and legal concerns. Federated learning allows collaborative model training while protecting raw healthcare records, but heterogeneous medical data distributions, limited feature representation, and insufficient secure and reliable model-sharing mechanisms among healthcare institutions often present challenges. This study proposes a Blockchain-Enabled Federated Learning Framework for privacy-preserving multi-disease prediction using the Federated Graph Multilayer Edge Attention Network (FGMEAN) and Great Wall Construction Algorithm (GWCA). Humboldt Squid Optimisation algorithm (HSOA) for optimal feature selection, Geometric Algebra Transformer (GATr) for better feature representation, FGMEAN for graph-aware federated disease prediction, and GWCA for adaptive hyperparameter optimization are used in the system. Blockchain and IPFS enable secure, decentralized, and verifiable model update and healthcare data interchange. The suggested architecture is tested on Heart Disease, Diabetes, and Chronic Kidney Disease datasets in federated healthcare. The experimental results show 99.3%, 99.5%, and 99.3% prediction accuracy for heart disease, diabetes, and chronic kidney illness. The framework also lowered latency, communication overhead, execution efficiency, and security compared to current techniques. Component-wise study shows that feature optimization, geometric feature embedding, graph-aware federated learning, and adaptive hyperparameter tuning improve predictive performance. Experimental results show that the BCFL-FGMEAN-GWCA framework provides a secure, privacy-preserving, and precise multi-disease prediction solution in distributed healthcare settings, enabling next-generation intelligent healthcare systems.

Keywords: Blockchain, Federated learning, Geometric Algebra Transformer, Great Wall Construction Algorithm, Humboldt Squid Optimization Algorithm.

3. User Evaluation of Conversational AI Systems: The Roles of Digital Literacy, Trust, and Perceived Privacy Risk

Authors and affiliations

• Guan Xian — School of Arts, Universiti Sains Malaysia, Malaysia

• Norfarizah Mohd Bakhi — School of Arts, Universiti Sains Malaysia, Malaysia

Abstract

Conversational AI is spreading rapidly, and understanding how users evaluate these systems has become an important issue in information systems research. Drawing on an information systems perspective, this study develops a structural model integrating digital literacy, trust, and perceived privacy risk to examine how digital literacy influences users’ evaluations of conversational AI systems through key psychological mechanisms. Using survey data collected from 580 Chinese Generation Z users, the study employs PLS-SEM to test the proposed model. The results indicate that digital literacy significantly enhances trust, which in turn positively influences system evaluation. Digital literacy is also positively associated with perceived privacy risk, suggesting that more digitally literate users are more sensitive to data practices and system uncertainty. However, perceived privacy risk does not significantly affect system evaluation, indicating that users tend to balance potential risks against perceived benefits during interaction. The study extends the application of Privacy Calculus Theory by identifying digital literacy as an important antecedent of user evaluation and highlighting the central role of trust in conversational AI contexts. Practically, the findings suggest that developers should enhance transparency, trust-building mechanisms, and privacy communication to improve user evaluations. Future research may further examine additional system-related factors and diverse user groups to validate and extend the proposed framework.

Keywords: Digital Literacy, Trust, Perceived Privacy Risk, Conversational AI, User Evaluation.

4. A Multi-Layer Contextual Model for Managing Data Overload in Smart Hydroponics

Authors and affiliations

• Ruziana Mohamad Rasli — School of Multimedia Technology and Communication, Universiti Utara Malaysia, Malaysia

• Sobihatun Nur Abdul Salam — School of Multimedia Technology and Communication, Universiti Utara Malaysia, Malaysia

Abstract

The widespread use of the Internet of Things (IoT) in hydroponic agriculture allows continuous monitoring of environmental conditions and provides real-time information to support crop management. However, the increasing amount of sensor data generated makes the monitoring process more complex as users need to interpret multiple environmental parameters simultaneously. Existing hydroponic monitoring systems focus on data collection and visualization, while the management and information delivery aspects to reduce data overload are still limited. This study proposes a multi-layer contextual model to manage environmental information in an intelligent hydroponic monitoring system. The model consists of five interconnected layers (Power, Sensor, Communication, Database and Visualization), which support the process of acquisition, validation, management and contextual data delivery. To validate the proposed model, a solar-powered IoT hydroponic monitoring system integrated with an Augmented Reality (AR) application was developed based on Situation Awareness Theory. Using System Usability Scale (SUS), the prototype was assessed by three domain experts. The results achieved an average usability score of 85.83, which reflects excellent usability. The results indicate that the proposed model improves organisation and presentation of environmental information by prioritising contextually relevant data rather than displaying all available sensor readings at once. This enables users to identify important environmental conditions efficiently and supports more effective monitoring. This approach helps to reduce data overload and provides a practical framework that can support the development of IoT-based smart agriculture applications in the future through the integration of anomaly detection techniques and predictive analytics.

Keywords: Data overload, Situation Awareness Theory, Multi-Layer Contextual Model, Internet of Things (IoT), Augmented Reality.

5. Computational Clickbait Detection: A PSALSAR-based Systematic Literature Review of Methodological Trajectories and Benchmarks (2016–2025)

Authors and affiliations

• Ari Wahyono — Faculty of Business and Information Technology, Universiti Muhammadiyah Malaysia, Malaysia; Faculty of Engineering, Universitas Tiga Serangkai, Indonesia

• Ku Ruhana Ku-Mahamud — Faculty of Business and Information Technology, Universiti Muhammadiyah Malaysia, Malaysia; School of Computing, Universiti Utara Malaysia, Malaysia

Abstract

Clickbait headlines have become a persistent issue in online media, where attention often takes priority over informational accuracy. While numerous computational approaches have been proposed to address this issue, the field remains fragmented across datasets, modelling strategies, and linguistic representations. This study presents a systematic literature review of clickbait detection research published between 2016 and 2025, following the Protocol, Search, Appraisal, Synthesis, Analysis, and Report framework to ensure a structured and reproducible review process. A total of 646 records were initially identified from Scopus, IEEE Xplore, Web of Science, and SpringerLink, from which 43 peer-reviewed journal articles were selected for in-depth analysis. Across these studies, a clear shift is evident from traditional feature-based machine learning toward deep learning and transformer-based models. Despite this progression, linguistic cues remain central, either through explicit feature engineering or implicitly learned representations. However, the review reveals a persistent gap: linguistic feature modelling and embedding-based approaches are often developed in isolation. To address this limitation, this study proposes a conceptual framework that integrates language-aware preprocessing, explicit linguistic features, and hybrid embedding strategies within a unified framework. By synthesising these components, the framework offers a more comprehensive perspective on clickbait detection, positioning it as an integrated linguistic and contextual modelling task rather than a purely classification problem. The findings and proposed framework provide a foundation for developing more robust, interpretable, and adaptable clickbait detection systems, particularly in multilingual and low-resource settings.

Keywords: PSALSAR Framework, Text Classification, Multimodal Processing, Transformer, Linguistic Feature.