Forthcoming Articles

International Journal of Web and Grid Services

International Journal of Web and Grid Services (IJWGS)

Forthcoming articles have been peer-reviewed and accepted for publication but are pending final changes, are not yet published and may not appear here in their final order of publication until they are assigned to issues. Therefore, the content conforms to our standards but the presentation (e.g. typesetting and proof-reading) is not necessarily up to the Inderscience standard. Additionally, titles, authors, abstracts and keywords may change before publication. Articles will not be published until the final proofs are validated by their authors.

Forthcoming articles must be purchased for the purposes of research, teaching and private study only. These articles can be cited using the expression "in press". For example: Smith, J. (in press). Article Title. Journal Title.

Articles marked with this shopping trolley icon are available for purchase - click on the icon to send an email request to purchase.

Online First articles are also listed here. Online First articles are fully citeable, complete with a DOI. They can be cited, read, and downloaded. Online First articles are published as Open Access (OA) articles to make the latest research available as early as possible.

Open AccessArticles marked with this Open Access icon are Online First articles. They are freely available and openly accessible to all without any restriction except the ones stated in their respective CC licenses.

Register for our alerting service, which notifies you by email when new issues are published online.

International Journal of Web and Grid Services (7 papers in press)

Regular Issues

  •   Free full-text access Open AccessA Novel Simplified Swarm Optimization of Convolutional Neural Network Weights
    ( Free Full-text Access ) CC-BY-NC-ND
    by Zhenyao Liu, Po-Hung Chiang, Chia-Ling Huang, Wei-Chang Yeh 
    Abstract: Convolutional Neural Network (CNN) is a feedforward neural network with multiple hidden layers using a neural network as the basic framework. CNN has the advantages of local neuron linkage and weight sharing and doesnt learn the same features under different filter processing, which improves computational efficiency and accuracy when avoiding over-fitting. In this study, we propose an algorithm based on Simplified Swarm Optimization (SSO), namely, SSO-CNN, as a solution to efficiently optimize the weights of one of the first released CNN models: LeNet. This method we developed using the advantages of SSO which is good at solving continuous problems. So that the accuracy of CNN can be improved stably in different test datasets. The results of experiments involving the MNIST, EMNIST, CIFAR-10, and SVHN datasets showed that the proposed algorithm was more efficient than traditional methods and it could be used with more complicated models like AlexNet, VGG, etc.
    Keywords: Convolutional Neural Network; Simplified Swarm Optimization; Machine Learning; Image Recognition; Weight Optimization.
    DOI: 10.1504/IJWGS.2026.10080880
     
  •   Free full-text access Open AccessRepurchase Prediction Modelling of Retail Customer Behaviour: an AI-Driven Approach
    ( Free Full-text Access ) CC-BY-NC-ND
    by Yu Yu Yen, Jui Hung Kao, Horng-Twu Liaw, Guang Lei Lu, Jui-Ying Chen 
    Abstract: This study investigates how artificial intelligencedriven analytics can be leveraged to predict repurchase behavior and enhance customer value management in automotive department store chains. Using an automotive retailer as an empirical case, transaction data from a POS system are analyzed through the RFM (Recency, Frequency, Monetary) model to segment customers by value. The results demonstrate that RFM-based segmentation effectively identifies heterogeneous customer groups with distinct repurchase intentions and purchasing patterns. Building on this segmentation, AI-enabled predictive modeling provides actionable insights for anticipating customer behavior, supporting personalized marketing strategies, and optimizing product and service portfolios. The findings underscore the importance of implementing differentiated retention strategies and utilizing data-driven decision support in competitive retail environments. By integrating customer analytics with AI-based prediction, this study offers both theoretical contributions and practical guidance for retailers seeking to strengthen customer loyalty and achieve sustainable profitability.
    Keywords: automotive industry; crm; rfm model; marketing strategy.
    DOI: 10.1504/IJWGS.2026.10081324
     
  •   Free full-text access Open AccessThe ADDIE Instructional Design Model combined with AI Technology in the Design of an Auxiliary Correction System for IP-Oriented Template-based Assignment
    ( Free Full-text Access ) CC-BY-NC-ND
    by Yung-Hui Chen, Lawrence Y. Deng, Chun-Hsiung Tseng, Jian-Yu Chen, Xiang-Yann Lim 
    Abstract: This study develops an Auxiliary Correction System (ACS) for IP-Oriented Template-based Assignments (IPOTA) in a vocational university network laboratory course. The instructional design of the course and the system implementation were guided by the ADDIE model to reduce grading workload while improving feedback timeliness and student learning support. Each assignment embeds the students seat number into the IP configuration so that every submission is unique, reducing the risk of copying but increasing grading complexity. To address this problem, the ACS combines rule-based text normalisation, an answer database, and image-based text extraction using deep-learning methods to assist grading. In the reported implementation, the ACS reduced the average grading time to about 20 seconds per assignment and achieved 100% accuracy on the training set for the image-recognition task. The results indicate that the proposed approach can substantially reduce manual marking effort and provide a basis for earlier instructional intervention and learning analytics.
    Keywords: ADDIE Instructional Design Model; Artificial Intelligence (AI); Auxiliary Correction System (ACS); IP-Oriented Template-based Assignment (IPOTA); Prediction Models.
    DOI: 10.1504/IJWGS.2026.10081396
     
  •   Free full-text access Open AccessAutomated medical report generation via two-stage trainable OCR and trustworthy generative AI
    ( Free Full-text Access ) CC-BY-NC-ND
    by Yu-Chih Wei, W.E.I. Shan LIN, Chia-Yun Hsu, Ching-Huang Lin 
    Abstract: As healthcare systems move toward networked architectures, the demand for trustworthy automated services grows. This study addresses manual reporting inefficiencies by proposing a framework integrating Two-stage Trainable OCR with the Trustworthy AI Dialogue Engine (TAIDE). The system utilizes CRAFT for text localization and YOLOv8 for symbol detection, followed by a CRNN-CTC module to handle complex ultrasound layouts. To ensure reliability and privacy, we incorporate Retrieval-Augmented Generation (RAG), a Natural Language Inference (NLI) mechanism for hallucination mitigation, and pixel-level de-identification. Experimental results show the architecture outperforms baseline OCR and increases the claim-level entailment ratio from 83.3% to 93.3%, a 10-percentage-point improvement in factual consistency. This solution offers a clinically applicable framework for AI-assisted reporting, significantly reducing manual workload while maintaining diagnostic safety and data privacy.
    Keywords: automated medical report generation; two-stage OCR; trustworthy AI; generative AI; natural language inference; hallucination detection.
    DOI: 10.1504/IJWGS.2026.10081735
     
  •   Free full-text access Open AccessRTTC: reliable trust-based topology control for secure wireless sensor networks
    ( Free Full-text Access ) CC-BY-NC-ND
    by Kun-Yu Lin, Sheng-Zhi Huang, Lin Hui, Mohd Yaseen Mir, Yi-Ting Wang, Chih-Lin Hu 
    Abstract: Conventional topology control algorithms for wireless sensor networks select cluster heads based solely on energy or probabilistic criteria, ignoring node trustworthiness. A compromised node that satisfies the energy criterion can therefore be elected to a pivotal cluster-head position, from which it silently discards aggregated data and blinds an entire network segment. This paper proposes a reliable trust-based topology control (RTTC) approach that integrates multi-dimensional trust evaluation directly into backbone formation. Each node is assessed along three dimensions, i.e., communication reliability modelled through subjective logic, residual energy availability, and social reputation derived from weighted neighbour recommendations, and the resulting composite score serves as the primary criterion for supernode selection. Simulations on both uniform-random and realistic OpenStreetMap topologies demonstrate that RTTC maintains delivery Copyright ? The Author(s) 2026. Published by Inderscience Publishers Ltd. This is an Open Access Article distributed under the CC BY license. (http://creativecommons.org/licenses/by/4.0/) 2 K-Y. Lin et al. ratios of 0.43 and 0.54, respectively, under 30% malicious-node conditions, whereas trust-agnostic baselines (LEACH, R-LEACH, OK-means) fall to 0.19-0.25 and 0.27-0.37 in the same scenarios.
    Keywords: trust management; topology control; network security; internal attacks; wireless sensor networks; WSNs.
    DOI: 10.1504/IJWGS.2026.10081736
     
  • ADAPT: Context-Aware Multi-Objective Scheduling Platform for Distributed Artificial Intelligence Services in the EdgeCloud Continuum   Order a copy of this article
    by Dhyonatan Santos De Freitas, Tathiana Duarte Do Amarante, Marcos Antonio Simplicio Junior, Mauricio Pillon 
    Abstract: Artificial intelligence as a service enables distributed AI deployment in Industry 4.0, yet placing services across the heterogeneous ECC infrastructures remains difficult: application requirements and infrastructure conditions change continuously, while existing solutions often rely on static, infrastructure-centric decisions. This paper presents ADAPT, a context-aware platform combining declarative service specification, a unified context model integrating application and infrastructure information, context-aware multi-objective scheduling, and technology-independent orchestration. ADAPT was validated in a real machining cell, running nine containerised AI services, including support vector machine (SVM), artificial neural networks (ANN), and convolutional neural network (CNN) workloads under edge-only, hybrid, and cloud-only configurations. Provisioning times averaged below 2 s in cloud and hybrid configurations, while Edge configurations reached observed peaks above 12 s. For a compute-intensive based in CNN workload, execution time was roughly 27 times higher at the edge than in the cloud. The hybrid arrangement exhibited execution times comparable to the cloud for some workloads while keeping latency-critical services near the data source, illustrating that service placement should adapt to application requirements and execution conditions.
    Keywords: Artificial Intelligence as a Service; Edge–Cloud Continuum; Context-Aware Scheduling; Multi-objective Scheduling; Industry 4.0.
    DOI: 10.1504/IJWGS.2026.10082191
     

Special Issue on: Emerging Technology in Big Data, AI, and Green Learning Driving Towards Sustainable Web Service Development

  • Optimality and Scalability of Semantic Web Service Composition with Hierarchical Parameter Relationship
    by Jung-Woon Yoo 
    Abstract: Semantic web service composition considers semantics for finding better solutions than syntactic web service composition. This paper focuses on hierarchical relationships among parameters of web services. A comprehensive mathematical model for semantic web service composition, into which hierarchical parameter relationships are incorporated, is presented as a general mathematical formulation. Experimental results demonstrate that the mathematical model for semantic composition finds hidden and better solutions that syntactic composition cannot find. The optimality of the solutions is empirically verified through extensive experiments. Furthermore, the scalability of the model is tested by comprehensive experiments to explore the impacts of eight key factors on web service composition. The mathematical model and the provided data sets are expected to serve as benchmarking tools for performance evaluation of heuristic algorithms for semantic web service composition. Finally, a web application is presented to visualize the semantic web service composition process, which is developed using the Django framework.
    Keywords: AI Planning; Web Service Composition; Semantics; Parameter Hierarchy; Mathematical Modeling.