Forthcoming Articles
International Journal of Information and Communication Technology
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 Articles will not be published until the final proofs are validated by their authors.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.
Articles 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 Information and Communication Technology (41 papers in press) Regular Issues
Abstract: To address the issues of high computational cost in the routing simulation of the field-programmable gate array physical design process and the neglect of the global topological structure of the circuit, a method for predicting the routability of integrated circuits based on deep learning is proposed. Circuits are abstracted into complex networks, from which topological features are extracted. These are fused with traditional circuit features and mapped onto heatmaps, achieving unified representation of geometric and topological information. Subsequently, the routing simulation task is transformed into an image generation problem. By introducing an efficient channel attention mechanism within a conditional generative adversarial network framework, the method highlights crucial feature channels to enhance generated image quality. Experimental results demonstrate that the proposed approach achieves 98.26% structural similarity and a peak signal-to-noise ratio of 41.75 dB, both significantly outperforming existing methods. This method provides a new paradigm for rapid assessment of routability prediction during the layout stage. Keywords: field-programmable gate array; FPGA; routing process simulation; attention mechanism; complex network; generative adversarial network; GAN. DOI: 10.1504/IJICT.2026.10079866
Abstract: To address the lack of systematic classification and trend analysis for creative packaging pattern design styles, this paper proposes a novel multimodal deep clustering framework. The method integrates convolutional neural network texture features, vision transformer global structure features, and simple linear iterative clustering balanced iterative reducing and clustering using hierarchies colour features, fusing them adaptively with learnable weights for dynamic feature fusion and end-to-end jointly optimised with deep embedding clustering to achieve unsupervised style classification. On the self-built packaging dataset containing 9,000 images, the method consistently achieves an accuracy of 0.768 and an F1-score of 0.755, thus significantly outperforming existing approaches (p < 0.05). The results reveal design evolution trends from 2015 to 2024, providing valuable data-driven support for design decision-making, trend prediction, and practical design practice. Keywords: clustering algorithm; packaging design; style classification; trend analysis; multimodal features. DOI: 10.1504/IJICT.2026.10079867
Abstract: This study tackles inefficiencies in virtual drama character action recognition. An improved skeleton extraction method employs sparse matrix operations and a novel activation function to reduce parameters and enhance nonlinearity. The core model uses an enhanced spatial-temporal graph convolutional network with a learnable matrix and multi-head graph attention to resolve topological rigidity. A twin network enables precise action quality assessment. Experiments demonstrate the superior performance. It achieves 97.8% average skeletal feature recognition accuracy across dramatic scenes and 95.85% accuracy in specific scenarios such as ballet jumps, outperforming comparable models. It also maintains robustness in noise, reaching 81.2% accuracy at a 0.25 noise level. For quality assessment, its evaluations align closely with expert judgements, achieving 98.6% average accuracy in rhythm matching tasks. In summary, this technique provides effective technical support for optimising virtual drama character action systems. Keywords: virtual drama; action recognition; ST-GCN model; attention mechanism; skeletal features. DOI: 10.1504/IJICT.2026.10079868
Abstract: Vocal emotional expression is a highly complex, time-varying process. Current generative models often struggle with long-term dependencies and fine-grained dynamic transitions, causing over-smoothed acoustic features. To address this limitation, a temporal-conditioned diffusion model is proposed for the high-fidelity simulation of time-varying vocal emotional features. The model utilises a stochastic differential equation based diffusion process operating on a multidimensional acoustic feature space. During the reverse generative phase, a hybrid LSTM-Attention network is employed to capture both macroscopic and microscopic emotional transitions. Comprehensive evaluations demonstrate that the proposed model significantly outperforms baselines. specifically, our model achieved a lower Mel-cepstral distortion (3.95 dB), reduced F0 RMSE (12.8 Hz), and a higher energy Pearson correlation (0.92). Furthermore, subjective mean opinion scores for acoustic naturalness (4.35) and emotional expressiveness (4.42) surpassed the best baseline. This study provides a robust generative paradigm for advanced affective computing. Keywords: vocal emotional characteristics; time-varying process; deep generative model; diffusion model; simulation modelling; emotional computing. DOI: 10.1504/IJICT.2026.10079869
Abstract: Academic performance prediction is essential for personalised teaching intervention, yet electromechanical courses involve complex precedence dependencies and logical correlations between knowledge points that traditional models fail to capture. To address this, we propose a learning-performance prediction method named knowledge-graph-based mechatronics education learning prediction, which integrates expert knowledge with historical interaction data to construct a knowledge graph encompassing 56 core knowledge points. This method employs graph convolutional networks (GCNs) to integrate topological features for representation learning and introduces a temporal-attention mechanism to model the dynamic evolution of knowledge mastery. Experimental results demonstrate that the proposed method achieves an area under the curve of 0.894, which is 5.2% to 11.7% higher than baseline models, and a normalised discounted cumulative gain at 5 of 0.856, significantly outperforming comparative methods. The findings highlight the effectiveness of incorporating knowledge-graph structures for accurate academic performance prediction in electromechanical education. Keywords: academic performance prediction; knowledge graph; graph convolutional network; GCN; temporal attention. DOI: 10.1504/IJICT.2026.10079870
Abstract: In complex retail systems, existing process models rely on static historical data, failing to capture the nonlinear dynamics of fluctuating consumer emotions throughout the market lifecycle. This paper proposes the emotion-decision synergistic simulation network, a novel simulation framework for lifecycle strategy optimisation. First, a deep extraction module using bidirectional long short-term memory and multi-head attention models the temporal evolution of emotional states from unstructured textual feedback. These continuous state vectors are then integrated into a multi-agent environment, explicitly simulating the dynamic state transitions and behavioural processes of virtual consumers under marketing stimuli. Extensive experiments across three large-scale datasets demonstrate that EDSSN significantly enhances process evaluation, reducing the mean absolute percentage error by up to 4.15% compared to traditional baselines. Ultimately, ours model provides a robust simulation testbed, successfully bridging qualitative human factors and quantitative process modelling for reliable decision support. Keywords: multi-agent reinforcement learning; bidirectional long short-term memory; Bi-LSTM; multi-head attention mechanism; demand forecasting; consumer sentiment. DOI: 10.1504/IJICT.2026.10079871
Abstract: Oil painting emotion recognition faces two challenges: the excessive number of model parameters makes deployment difficult, and the granularity of emotion semantic understanding is coarse. To address this issue, this paper proposes a lightweight representation learning method that integrates knowledge distillation and multi-granularity semantic alignment. The teacher network guides the lightweight student network to learn, and performs feature alignment at two granularities: global image semantics and local emotional attributes, enabling the compact model to accurately capture the subtle emotions in the paintings while maintaining efficiency. Experiments show that, with a 84.7% reduction in parameters, the classification accuracy only drops by 1.2%, and the normalised discounted cumulative gain for emotion intensity ranking reaches 0.893, significantly outperforming the lightweight baseline models of the same scale. This research provides a technical solution that balances efficiency and accuracy for the practical implementation of emotion computing in the field of digital humanities. Keywords: knowledge distillation; KD; multi-granularity semantic alignment; oil painting sentiment analysis; sentiment computing. DOI: 10.1504/IJICT.2026.10079872
Abstract: Financial statement fraud often hides behind locally plausible numbers while breaking cross-statement coherence, which weakens traditional ratio screens and single-source detectors. To address this challenge, a dual-layer knowledge graph and graph-driven fraud detection framework is proposed. In the framework, first, accounting semantics and consistency constraints are organized as a schema layer and linked to firm-year observations as an instance layer. Then, layer-coupled graph learning aligns observations with accounting meaning and extracts compact evidence subgraphs for audit review. Finally, a generative consistency module estimates coherent reporting profiles to expose distributed manipulation through residual inconsistency. Experimental results show that the proposed method achieves an average precision of 0.74 and improves low-false-alarm recall to 0.60, outperforming representative baselines by up to 0.14 in average precision. The framework provides reliable risk ranking and practical interpretability for audit-oriented deployment. Keywords: financial statement fraud; dual-layer knowledge graph; heterogeneous graph learning; evidence subgraph extraction. DOI: 10.1504/IJICT.2026.10079873
Abstract: Supply and guarantee chains among enterprises create complex networks with significant risk transmission effects. Traditional early warning models relying on isolated financial indicators cannot capture this dynamic evolution, causing delayed and inaccurate warnings. This paper proposes temporal knowledge graph dynamic graph neural network, a real-time framework integrating temporal knowledge graphs and dynamic graph neural networks. It constructs a multi-relation temporal knowledge graph containing enterprises, transactions, and guarantees to depict risk transmission paths, and designs a time-aware graph attention mechanism. Experiments on public datasets show the proposed method achieves 0.962 area under the curve, 3.2% higher than graphshield, and 0.781 normalised discounted cumulative gain at position 10, 8.7% higher than dynamic graph neural network with static relationships. Results demonstrate this technology effectively identifies early risk signals in network environments, providing timely and interpretable decision support for business and finance risk control. Keywords: temporal knowledge graph; dynamic graph neural network; business and finance risk; early warning. DOI: 10.1504/IJICT.2026.10079874
Abstract: Traditional assessments of student legal literacy rely on static questionnaires or performance analysis, which often fail to capture structural complexity and predict highrisk behaviours accurately. To address this, we propose an integrated algorithm combining logistic regression (LR) and a latent structural model (LSM) based on the twoparameter logistic IRT model. LSM estimates latent traits across legal knowledge, concepts, and abilities, mitigating nonlinearity and measurement error, while LR incorporates demographic, behavioural, and latent trait data to predict future highrisk behaviour. A joint optimisation mechanism quantifies literacy and enables early warning of risks. Experiments show the fusion model achieves 88.5% accuracy in highrisk behaviour prediction, outperforming standalone LR (78.2%) and structureddataonly models (75.5%). It also yields an F1 score of 0.875 and recall of 86.0%, demonstrating robust performance. The approach offers a refined framework for legal literacy assessment and supports targeted prevention strategies in educational management. Keywords: legal literacy assessment; logistic regression; latent structural model; LSM; latent traits; high-risk prediction. DOI: 10.1504/IJICT.2026.10079875
Abstract: This study addresses the mismatch between proliferating multimodal resources and learners cognitive needs in digital education. Conventional manual annotation and static classification fail to capture cross-modal semantic associations or adapt to evolving cognitive levels. We propose a knowledge forest-transformer model integrating hierarchical knowledge organisation with self-attention mechanisms for multimodal semantic association and cognitive adaptation. Structured multimodal feature modelling improves cross-modal accuracy: text-image from 68.2% to 89.7%, text-audio from 59.8% to 83.6%, and image-video to 90.2%. Cognitive adaptation achieves 85.8% success at threshold 1.8, while cognitive labels enhance F1-scores (analysis layer: 61.2% to 79.5%; application layer: +17.8%). Response times under three seconds yield cognitive load of 4.2 (12.5% exceedance) versus 8.4 (68% exceedance) when exceeding eight seconds. After four weeks, low-ability learners improve mastery from 32.1% to 61.3%, with medium and high-ability groups reaching 82.7% and 94.2%, respectively. Results demonstrate effective multimodal resource association and cognitive adaptive support. Keywords: multimodal teaching resources; semantic association; cognitive level adaptation; knowledge forest; transformer. DOI: 10.1504/IJICT.2026.10079876
Abstract: To address the challenges of structural coherence and stylistic encoding in ethnic music generation, this paper proposes a novel three-dimensional sequence convolutional network (3D-SCN). The model integrates the translational invariance of CNNs, the sequence modelling of bi-LSTMs, and the feature fusion of multi-head attention. Through multi-dimensional feature extraction and spatiotemporal modelling, it captures patterns in traditional Chinese pentatonic music. Experiments show the model outperforms baselines like music transformer, achieving 98.7% accuracy, a 0.941 F1-score, and 95.6 in expert-rated ethnicity. It also demonstrates strong cross-cultural adaptability (91.3% cultural accuracy) and long-sequence generation capability (92.8% structural integrity). Despite computational complexity, this work provides an effective framework for digital music generation and heritage. Future work will focus on model lightweighting, dynamic memory integration multi-voice expansion. Keywords: three-dimensional sequence; convolutional network; folk music; stylisation. DOI: 10.1504/IJICT.2026.10079877
Abstract: This study examines the incorporation of artificial intelligence into intelligent institutions, focusing on the lack of a cohesive closed-loop architecture that integrates multimodal data intelligence, decision analytics, and policy optimisation. The AI-orchestrated digital transformation framework (AI-ODTF) has three levels: data intelligence, decision intelligence, and governance optimisation. multimodal data, relational embeddings, and transformer-based temporal encoding are used to make a single institutional knowledge graph. Ensemble prediction and Bayesian causal inference make it possible to evaluate policies that do not happen, while reinforcement learning with multi-objective optimisation finds the best balance between quality, cost, and sustainability. The framework is confirmed utilising the xAPI-Edu-Data dataset, employing an 80:20 train-test division and five-fold cross-validation. AUC, RMSE, F1-score, and counterfactual gain are all used to measure performance. The results show big improvements, such as an AUC increase from 0.74 to 0.91, a 19.6% increase in resource efficiency, and a 17.9% increase in overall performance. Keywords: artificial intelligence; digital transformation framework; smart universities; data intelligence layer; decision intelligence layer; governance optimisation layer. DOI: 10.1504/IJICT.2026.10079878
Abstract: The use of technology in the learning and medical fields has significantly transformed the student participation particularly in physical education. College students often find it difficult to exercise regularly and are therefore less physically fit and achieve poor grades. This research paper will look at how wearable devices impact physical education attendance and academic achievement amongst college students who take physical education classes. A new system called the wearable activity monitoring and engagement system (WAMES) unites wearable fitness trackers along with a structured exercise program and real-time feedback systems. During a 12 week semester, we monitored the frequency, duration, intensity and performance of the participants in school. The findings indicate that students who consumed WAMES had 35 % higher chances of exercising, scoring higher in school, and having improved physical performance, which showed that they were more motivated, adherent, and had better learning results. Keywords: wearable devices; exercise engagement; academic performance; physical education; PE; student motivation; technology-enhanced learning. DOI: 10.1504/IJICT.2026.10079972
Abstract: Cross-border e-commerce warehousing and logistics face severe challenges of dynamic and variable demands as well as complex networks. Traditional static planning methods are unable to achieve real-time optimisation. Therefore, this paper proposes a spatio-temporal graph neural network model, which can simultaneously capture the spatial correlations between warehouses and transportation routes in the logistics network and the changing patterns of order flows over time, thereby enabling dynamic and intelligent allocation of warehousing resources. The experiments are based on real cross-border transaction data, and the results show that compared with the optimal benchmark model, this method increases the order fulfilment rate by 8.2% and reduces the average logistics cost by 15.1%. The research indicates that this method provides an effective approach for building an efficient and flexible intelligent supply chain system. Keywords: spatio-temporal graph neural network; STGNN; cross-border e-commerce; warehousing logistics; dynamic configuration. DOI: 10.1504/IJICT.2026.10079973
Abstract: In the hand-painted deduction in the early stage of environmental design, the 2D concept sketch is easy to cause the 3D space skeleton clues contained in the bottom layer to be extremely chaotic and difficult to analyse. This paper proposes a self-supervised learning framework based on geometric consistency, which is used to automatically extract perspective cues directly from environmental design hand-drawn drafts. The technical mechanism proposed in this paper can directly utilise the inherent orthogonal patterns and vanishing point distribution within the drawings to autonomously establish spatial constraints. Experiments show that the framework proposed in this paper achieves an average accuracy rate of 88.5% in extracting perspective cues, achieving a 4.2% performance improvement compared to the current mainstream benchmark methods, and controlling the parsing time of a single drawing within 1.2 seconds. This method effectively bridges the technical gap between conceptual hand-drawn art and precise computer-aided modelling. Keywords: environmental design; perspective cues; self-supervised learning; sketch analysis; intelligent design assistance. DOI: 10.1504/IJICT.2026.10079974
Abstract: This study proposes a physics-informed neural network (PINN)-based edge-cloud collaborative digital twin (DT) for predictive maintenance of aging railway signalling equipment (RSE). The core model, an adaptive pigeon-inspired optimised PINN (API-PINN), integrates electromechanical degradation physics with operational data to forecast faults and estimate remaining useful life (RUL). The architecture enables low-latency, real-time anomaly detection at the network edge while leveraging the cloud for intensive model training. Experiments on real-world railway signalling data sets can show that API-PINN attains the prediction accuracy of 98.77% and the R2 of 0.968, which is better than the baseline models. The given framework offers a strong performance when the data is noisy and imbalanced, allowing for the arrangement of the maintenance schedule proactively, minimising the costs of operations, minimising downtimes, and improving safety. Keywords: predictive maintenance; digital twin; edge-cloud computing; railway signalling equipment; RSE; aging infrastructure. DOI: 10.1504/IJICT.2026.10079990
Abstract: The total ventilation rate of cigarettes is an important physical parameter of cigarette products. The stability of the ventilation rate has a significant impact on the sensory quality of cigarettes and the stability of mainstream smoke. This study takes a medium-sized cigarette brand as the research object and constructs a prediction model for the total ventilation rate of cigarettes using the XGBoost algorithm. The interpretation and analysis of the model using the SHAP method indicate that, without changing the quality indicators of cigarettes, the stability of the ventilation rate can be improved by adjusting the actual length of the ventilation band of the mouthpiece paper, the starting position of the ventilation band of the mouthpiece paper, and the set value of the compaction end position. These adjustments can enhance the control level of the stability of the ventilation rate of cigarettes and guide the management and operational practices of ventilation rate indicators in the production workshops. Keywords: XGBoost; SHAP value; total ventilation rate; process quality. DOI: 10.1504/IJICT.2026.10080033
Abstract: The paper proposes a load optimisation adaptive algorithm DKR LOAD) for dynamic knowledge retrieval enhancement. Through innovative fusion of retrieval enhancement generation (RAG) and LLMfine-tuning technology, it introduces a dynamic knowledge retrieval module and a weight adjustment model based on self-criticism preference optimisation (SCPO) to achieve real-time knowledge update and parameter optimisation. The experimental results show that on the lEEE PES dataset, the algorithm has an accuracy rate of 92.0%,an Fl score of 0.9l, and a response time of only 600 ms, which is better than baselines such as GPT-3 (82.5% accuracy). In the robustness test, the noise variance of the algorithm is as low as 0.01, and the practicality expert score is 4.5 points, which proves that it can effectively semantic understanding accuracy, anti-interference ability and practical application value. It offers a scalable solution for power communication. Keywords: power communication; large language models; LLM fine-tuning; retrieval enhancement generation; RAG; dynamic knowledge retrieval; weight optimisation. DOI: 10.1504/IJICT.2026.10080034
Abstract: To predict users purchase intention, in terms of feature extraction method, an improved graph sampling aggregation algorithm that introduces time series and location information is designed. Next, the study proposes a user purchase intention prediction algorithm that combines graph sampling aggregation algorithm and higher-order graph neural network. The experimental results based on the public dataset of mainstream e-commerce platforms from 2022 to 2024 show that, the feature extraction capability of the improved graph sampling aggregation algorithm was significantly better than the traditional method. The performance of the willingness prediction algorithm was even more outstanding. In experiments comparing it with high-order graph neural networks, lightweight gradient boosting machines, graph convolutional network logistic regression combination models, etc., its accuracy was above 92%, the maximum F1 value was 0.967. The performance of the feature extraction method and prediction algorithm is better than that of the comparison model. Keywords: purchase intention; prediction; GraphSAGE; HOGNN; electric commerce; e-commerce. DOI: 10.1504/IJICT.2026.10080035
Abstract: In online education, students home lighting conditions are variable, and traditional methods suffer from a decline in facial feature extraction accuracy under low light conditions. This paper proposes a face detection model that integrates local binary pattern texture features with FaceNet deep embedding features. Based on multi-task cascaded convolutional networks, low light texture perception is enhanced through local binary patterns, identity consistency constraints are provided through FaceNet, and deep feature fusion is performed through a 50 layer residual network to improve the robustness of keypoint detection in dim environments. In addition, the eye aspect ratio and mouth aspect ratio are calculated, and a three-path head pose estimation algorithm is integrated to achieve multidimensional quantitative analysis of the learning state. Experiments show that the key point detection accuracy under extremely low lighting conditions reaches 95.03%, with an average error of 4.12 pixels. The recognition accuracy for fatigue and other states in the educational dataset exceeds 95%. The classroom video analysis successfully plotted the time series variation curves of different students attention levels. The contribution of this study lies in proposing a face detection model for low-light environments and constructing a multi-dimensional feature system that integrates facial expressions and head poses, and verifying the effectiveness of this solution in the online education scenario. Keywords: LFM-FD; facial expression recognition; learning state analysis; convolutional neural network; feature extraction. DOI: 10.1504/IJICT.2026.10080036
Abstract: The study proposes to improve the OpenPose estimation algorithms skeleton extraction algorithm, which is based on microgrid and merged convolution, to address the issue that the current methods for assessing the quality of taekwondo movement are not accurate or valid enough to meet practical needs. Based on this improved algorithm, an algorithm is constructed to assess the quality of taekwondo movement. Experimental results show that the optimised OpenPose algorithm achieves 92.8% accuracy, 95.2% recall, and a validation loss value of 0.72, demonstrating better overall performance than the comparison methods. The researchs suggested OpenPose skeleton extraction algorithm and action quality evaluation algorithm outperform the comparison algorithm in regard to accuracy, performance, and running speed, and have useful applications. Keywords: OpenPose; skeleton extraction; taekwondo; sport quality assessment; intelligence. DOI: 10.1504/IJICT.2026.10080074
Abstract: In the quality evaluation of innovation and entrepreneurship education, redundancy of the original indicator system often leads to ambiguous clustering boundaries, failing to distinguish substantial differences among universities. To address this, this paper adopts mutual information screening to select the feature subset most correlated with educational quality, then conducts cluster analysis on universities. Experiments on 2,100 student entrepreneurial projects show that after mutual information screening, normalised mutual information increases from 0.623 to 0.847, the Davies-Bouldin index decreases from 1.052 to 0.738, and inter-group differences of two domain-specific indicators (entrepreneurial intention retention rate and industryacademia integration tightness) expand by 38.6% and 42.3%, respectively. This framework reveals that innovation score, funding amount, and team experience are core factors driving educational quality differentiation, providing a quantifiable classification basis for educational management departments. Keywords: mutual information; feature selection; innovation and entrepreneurship education; clustering evaluation. DOI: 10.1504/IJICT.2026.10080075
Abstract: To solve the problem of low precision of power system fault diagnosis, a transformer fault diagnosis model for power system based on improved CNN-BiLSTM and multimodal feature fusion is proposed. Firstly, this paper adopts improved CNN to extract infrared image features of power system transformer, and uses BiLSTM network to extract oil chromatograms and local discharge sequence features of power system transformer. Moreover, multimodal feature fusion algorithm is utilised for feature fusion and fault diagnosis. Finally, simulation experiments are conducted for verification. Experimental results show that the average false detection rate and average loss detection rate of proposed model are 2.99% and 3.53%, respectively, which are about 6% lower than those of other commonly used fault diagnosis models. Therefore, the proposed model improves the precision of early warning and fault diagnosis for power system transformers, providing technical support for ensuring the normal operation of the power system. Keywords: power system; transformer; fault diagnosis; CNN network; BiLSTM network; multi-modal feature fusion algorithm. DOI: 10.1504/IJICT.2026.10080077
Abstract: In nursing simulation training, student performance changes dynamically as they acquire skills. Traditional knowledge tracking can only provide passive assessments and cannot actively adjust strategies, resulting in a disconnect between prediction and intervention. This paper proposes an adaptive prediction framework that integrates dynamic knowledge tracking with quality-function learning. A recurrent neural network tracks the temporal evolution of mastery of nursing knowledge points, using learning gain maximisation as a reward, while an agent dynamically optimises the difficulty and sequence of practice tasks. Using a publicly available nursing skills interaction dataset, this framework improved the area under the curve from 0.790 to 0.873 and the accuracy from 0.742 to 0.801. Comparative analysis demonstrates that integrating reinforcement learning with knowledge tracking effectively enhances predictive performance, providing interpretable insights for personalised interventions in nursing simulation education. Keywords: nursing education; adaptive reinforcement learning; student performance prediction. DOI: 10.1504/IJICT.2026.10080079
Abstract: During the building operation and maintenance stage, the factors influencing carbon emissions are complex and highly coupled. The existing diagnostic methods based on correlation analysis or black-box models are unable to reveal the underlying causal mechanisms. This paper integrates the static topology and equipment attributes in the building information model as prior knowledge into a constrained causal discovery framework, learns causal structures from the dynamic data of the internet of things, and realises the diagnosis from correlation descriptions to causal tracing. Experiments on public datasets show that the accuracy rate, recall rate, and area under the curve value of this method in identifying key influencing factors are 0.92, 0.88, and 0.95 respectively, which are 14%, 11%, and 7% higher than the standard Peter-Clark algorithm, and can track key causal chains such as outside temperature, air conditioning set temperature, carbon emissions. Keywords: building information model; causal discovery; carbon emission; traceability diagnosis. DOI: 10.1504/IJICT.2026.10080097
Abstract: The speed at which digital technology advances and the need for low-carbon development make coordinated policies all the more relevant for sustainable development. This paper introduces a comprehensive analytical framework that merges text analysis with time series modelling to methodically examine the relationship between digital and environmental policies. By converting policy documents into quantitative indicators, a more accurate and dynamic evaluation of their efficacy becomes possible, surpassing shortcomings of conventional qualitative approaches. The research uses provincial data to analyse the long-term impacts of synchronised policy implementation on economic growth, ecological preservation, and technological progress. Better coordination leads to greater productivity, less pollution, and faster innovation, though these effects differ greatly across regions. The key transmission channels are industrial upgrading and more efficient resource use. The research offers a practical and clear method to comprehend policy interactions and improve policy design. Keywords: digital development; policy coordination; text analysis; time series methods; data-based research. DOI: 10.1504/IJICT.2026.10080098
Abstract: In online learning scenarios, the complex dependencies between knowledge concepts are often simplified, and the dynamic changes in learners cognitive states are difficult to capture. As a result, personalised learning path recommendations generally lack the capacity to achieve global optimisation. To address this, the paper proposes a knowledge graph-enhanced reinforcement learning approach for personalised learning path optimisation. This method constructs a heterogeneous knowledge graph and introduces a graph attention network (GAT) to enhance concept representation, then employs an actor-critic agent to learn optimal decision strategies under composite reward guidance. On the MOOCCubeX dataset, the method improves Precision@5 from 0.504 to 0.547 and normalised discounted cumulative gain@5 from 0.489 to 0.535; on the MOOPer dataset, Precision@5 improves from 0.462 to 0.508, and normalised discounted cumulative gain@5 improves from 0.441 to 0.488. Keywords: knowledge graph; reinforcement learning; personalised learning paths; agents. DOI: 10.1504/IJICT.2026.10080099
Abstract: In competitive sports, sources of stress for athletes such as critical moments in competition, audience distractions, and coaches expectations vary greatly from person to person. Traditional questionnaire-based methods are subjective and lag behind events, making it difficult to support real-time interventions. Single recommendation algorithms, lacking prior knowledge of stressor types, tend to produce one-size-fits-all strategies, which can exacerbate mental fatigue. To address this challenge, this paper proposes an identification framework that integrates hierarchical clustering with hybrid collaborative filtering: first, typical stressor patterns are extracted through clustering, and then personalised stress relief strategies are generated based on these patterns. Validation using publicly available physiological stress datasets shows that, compared to traditional collaborative filtering, our method improves accuracy from 78.3% to 86.7%, increases the area under the curve from 0.83 to 0.89, and raises the normalised discounted cumulative gain @10 from 0.72 to 0.81. Keywords: identification of competitive stressors; cluster analysis; collaborative filtering; personalised recommendations. DOI: 10.1504/IJICT.2026.10080100
Abstract: In the emotional understanding of childrens stage narratives, childrens emotional expressions are often embedded within the narrative flow of the performance. However, existing methods focus on isolated facial expressions or speech features, ignoring the dynamic evolution of emotions over time and their significant migration within the narrative context. This paper proposes the adaptive attention flow network, which introduces an attention flow field based on conservative flux constraints to dynamically track key emotional nodes across time, space, and narrative. The attention weights are regarded as information flows that can be transferred across frames, ensuring smooth evolution and lossless transmission of emotional representations along the narrative timeline. Experiments show that the classification accuracy of the adaptive attention flow network reaches 84.6% with an area under the curve of 0.91, outperforming the timesformer baseline (74.2% accuracy and 0.83 area under the curve) by 10.4 percentage points and 0.08, respectively. Keywords: emotional computing; children's stage narrative; attention flow; multimodal fusion. DOI: 10.1504/IJICT.2026.10080101
Abstract: To address the limitations of traditional large language models in adapting to the intelligent construction of new power systems, as well as the insufficient recall of domain-specific knowledge, semantic associations, and topological relationships in conventional retrieval-augmented generation (RAG) architectures, this paper constructs a knowledge graph-enhanced RAG system for the power domain. It proposes optimisation methods including domain-adaptive semantic embedding, knowledge graph-enhanced multi-strategy retrieval, and dual-constrained compliance generation. Comparative experiments with traditional RAG models demonstrate that the proposed model outperforms conventional RAG in retrieval precision and hallucination suppression, effectively meeting the requirements of various core business scenarios in new power systems and demonstrating strong engineering value. Keywords: retrieval-augmented generation; RAG; power knowledge graph; large language model; LLM; power system intelligence. DOI: 10.1504/IJICT.2026.10080102
Abstract: In industry-education integration scenarios, existing knowledge tracking models are largely based on statistical correlations derived from historical interactions, making it difficult to distinguish between skill deficiencies and random errors; moreover, their black-box nature undermines the credibility of decision-making. This paper proposes a causality-enhanced attention mechanism that models causal pathways between skill mastery, exercise interactions, and job requirements using structural causal graphs. By introducing counterfactual reasoning to generate representations of knowledge states after interventions, the model focuses on actual changes in skills rather than superficial patterns in interaction sequences. Experiments on the ASSISTments and industry-education integration platform datasets demonstrate that the model achieves area under the curves of 0.834 and 0.871, respectively, representing improvements of 3.1% and 2.7% over the best baseline. normalised discounted cumulative gain @5 also significantly outperforms comparison methods, and the visualisation of causal attention weights reveals the models precise attribution of skill-related root causes. Keywords: knowledge tracking; causal inference; industry-academia integration; interpretability; attention mechanisms. DOI: 10.1504/IJICT.2026.10080103
Abstract: In order to solve the frequent loss of small symbols in complex music scores and the serious interference of five line staff backgrounds, a research proposes an end-to-end architecture that integrates music score recognition and style generation. The recognition stage adopts a multi-scale feature fusion convolutional recurrent neural network, combined with Haar wavelet transform to preserve edge details, and utilises note focusing module and quasi recurrent neural network for efficient transcription. The generation stage adopts a hierarchical cascade strategy, combining generative adversarial networks and Transformers to capture structural and spectral details, and then synthesising waveforms through WaveNet voice encoders. The results showed that the symbol accuracy on the DeepScores V2 dataset was 96.82%, and the symbol error rate under noise interference was only 5.15%. The research results provide effective technical support for the audio-visual conversion from visual music scores to auditory audio. Keywords: convolutional recurrent neural network; CRNN; music score recognition; generative adversarial network; GAN; transformer; music generation. DOI: 10.1504/IJICT.2026.10080147
Abstract: The digital trade flow patterns within manufacturing clusters are crucial for analysing supply chain resilience. However, existing graph neural network methods often treat enterprises within the cluster as homogeneous nodes, ignoring structural differences such as product specialisation levels and technical density, resulting in significant deviations in trade intensity predictions. This paper proposes the domain-aware attention graph network, which explicitly captures the heterogeneous trade relationships among enterprises by embedding learnable node domain feature modulation functions. Experiments show that domain-aware attention graph network achieves a determination coefficient of 0.842 in the trade intensity regression task, an improvement of approximately 0.049 compared to graph attention network; in the trade change direction classification task, its area under the curve reaches 0.891, an improvement of approximately 0.044 compared to graph attention network, indicating that the model can more accurately identify trade flow and intensity changes. Keywords: manufacturing cluster; digital trade flow; hierarchical attention mechanism; gravitational model. DOI: 10.1504/IJICT.2026.10080148
Abstract: In order to meet the demand for the digital modelling and intelligent analysis of students learning behaviour, a predictive evaluation framework is proposed, which combines feature engineering with LightGBM. Through the construction of tags and the extraction of training samples from college English course behaviour log, the construction strategy and interpretable output mechanism of the model are discussed. The results show that the proposed model achieves an F1-score of 0.889 on the test set, with an average response latency of less than 230 ms, and a performance fluctuation of less than 1.6% in 5 times cross validation, demonstrating robust deployment stability and generalisation. Keywords: learning behaviour prediction; LightGBM algorithm; educational data mining; Intelligent teaching analytics; learning behaviour modelling. DOI: 10.1504/IJICT.2027.10080149
Abstract: Accurate prediction of copper ore mineralisation potential remains a challenging task in mineral resource exploration. To address the limitations of existing methods in characterising complex mineralisation processes and their unsatisfactory predictive performance, this paper proposes a novel deep learning framework with three key innovations. First, multi-source geological data are pre-processed using an entropy-weighted principal component analysis strategy to remove redundant features. Second, a meanshift-improved synthetic minority over-sampling technique is applied to oversample the training set, effectively balancing positive and negative sample ratios. Third, asymmetric convolution modules and a multi-spectral channel attention module are integrated into the ResNeXt50 network, significantly enhancing the model's ability to extract features from small-sized, multi-channel mineralisation data. Experimental results demonstrate that the proposed method achieves a prediction accuracy of 95.14%, outperforming the best benchmark method by at least 3.03%, and enables efficient simulation of copper ore mineralisation. Keywords: multi-source geological data; deep learning; synthetic minority over-sampling technique. DOI: 10.1504/IJICT.2026.10080150
Abstract: Personalised English vocabulary recommendation remains difficult because learner behaviour changes quickly while lexical relations are highly structured. To address this issue, this study proposes an attention-enhanced knowledge graph recommendation model for vocabulary learning. First, learner interaction sequences are encoded to capture recent review rhythm, repeated errors, and short-term preference shifts. Then, lexical relations are organised through a vocabulary knowledge graph to preserve semantic, derivational, and pedagogical connections among words. Finally, both sides are fused in a personalised scoring stage to generate the next-word recommendation list. Experimental results show that the proposed method achieved 0.742 Precision, 0.694 Recall, 0.781 NDCG, and 0.726 MRR on Dataset A, and remained consistently superior on Dataset B. The model shows strong ranking accuracy, stable recommendation behaviour, and good educational suitability. Keywords: personalised vocabulary recommendation; knowledge graph; attention mechanism; English learning. DOI: 10.1504/IJICT.2026.10080151
Abstract: Traditional studies on bel canto resonance rely mainly on subjective auditory evaluation, lacking standardised quantitative methods. It is difficult to accurately capture the internal correlations and variation patterns of multi-scale resonance characteristics, it also makes the optimisation of bel canto teaching and singing skills lack of scientific quantitative basis. To address this issue, this paper proposes a dedicated multi-scale feature extraction algorithm and an attention-based feature fusion algorithm. Combined with acoustic experiments on soprano singing signals, we conduct a systematic quantitative study. Results show that this method can effectively extract multi-scale features of bel canto resonance. The recognition rate of the fused features for bel canto vocal types is significantly higher than that of the single feature, and the average recognition rate of all vowels is more than 90%, which can accurately reveal the characteristic distribution of bel canto resonance. Keywords: bel canto; resonance characteristics; multiscale feature; acoustic quantitative analysis. DOI: 10.1504/IJICT.2026.10080152
Abstract: As engineering ethics teaching faces the contradiction between overload of case resources and increasing demand for individualisation, this study aims to develop a system that can provide accurate and explainable recommendations. By constructing the KGDP-RM model, which integrates knowledge graphs with differential privacy protection, we employ graph neural networks to learn case semantic association, and attention mechanism and Bayesian network are introduced to realise dynamic preference capture and causal explanation. The accuracy rate of the model on the EdNet dataset reaches 0.821, the F1-value is 0.813, representing an improvement of 0.038 over the best-performing baseline. Its performance degrades by only 0.024 in a 10% noise environment. The user relevance and satisfaction scores reach 4.3 and 4.1 points respectively. In addition, the ablation test confirms that the knowledge graph module contributes the most. Overall, the multi-technology integration framework proposed in this study effectively solves the problems of sparsity, privacy security and interpretability in teaching resource recommendation. Keywords: engineering ethics case teaching; personalised recommendations; knowledge graph; differential privacy; interpretability. DOI: 10.1504/IJICT.2026.10080173
Abstract: With the vigorous development of the global experience economy, the problem of overloading tourist traffic in popular scenic areas has become increasingly prominent, posing a huge threat to public safety. The traditional centralised cloud computing monitoring architecture, when dealing with massive sensing data, has bottlenecks such as communication congestion, high end-to-end latency, and high energy consumption at the edge nodes. This study proposes a communication-efficient distributed sensor network framework (CEDF), adopting a three-layer collaborative architecture of edge-fog-cloud, and downloading the feature extraction task to the edge nodes, while introducing an event-triggered communication protocol based on adaptive dynamic thresholds and a distributed state consensus algorithm. In a real scenic area with an area of approximately 25 square kilometres, 500 heterogeneous sensor nodes were deployed for on-site verification. The results showed that compared with the traditional continuous periodic transmission (CPT) and standard edge clustering (SEC) architectures, CEDF reduced the total network data communication overhead by over 90%, compressed the end-to-end closed-loop control delay to within 30 milliseconds, and increased the average tourist circulation efficiency by 34%. This research provides an engineering paradigm for the large-scale deployment of smart tourism systems. Keywords: distributed sensor network; event-triggered communication; real-time tourist flow monitoring; edge computing; smart tourism; congestion control; standard edge clustering; SEC; continuous periodic transmission; CPT. DOI: 10.1504/IJICT.2026.10080257
Abstract: Cross-modal hashing facilitates efficient retrieval across diverse modalities by mapping shared semantics into a compact binary space. Despite significant progress, conventional approaches frequently encounter two bottlenecks: 1) a preoccupation with Euclidean geometry that overlooks intrinsic graph structures, making them susceptible to data noise/outliers; 2) a reliance on a single latent subspace which fails to capture the complexity of multi-modal distributions. To circumvent these limitations, we introduce refining crossing modal retrieval via two stage graph hashing (RCRT). In this initial phase, collective matrix factorisation is integrated with a customised graph convolutional network (GCN) to extract modality-specific latent representations. By merging Euclidean features with non-Euclidean structural priors, this stage robustly mitigates the impact of anomalous data points. Subsequently, a global approximation strategy aligns these heterogeneous subspaces to distil high-level common semantics, followed by the derivation of discrete hash codes that preserve semantic consistency. In the second phase, a linear hash function is optimised via a local similarity preservation mechanism, ensuring the fine-grained topology is maintained within the Hamming space. Across multiple benchmark datasets, RCRT demonstrates superior performance over all current SOTA methods. Keywords: discrete optimisation; crossing modal retrieval hashing; graph convolutional. DOI: 10.1504/IJICT.2026.10080279 |
Open Access
Keep up-to-date
-
Our Blog -
Join us on Bluesky -
Follow us on X -
Visit us on Facebook - Our Newsletter (subscribe for free)
- RSS Feeds
- New issue alerts
- Inderscience is a member of publishing organisations including:
-
-
