International Journal of Information and Communication Technology (10 papers in press)

Regular Issues

• Open Access

  Early warning of college students ideological public opinion based on TF-IDF and RFB neural network ( Free Full-text Access )

  by Guixue Tan 
  Abstract: To address the need for real-time early warning of college students social media opinions, this study proposes a dynamic model integrating term frequency-inverse document frequency (TF-IDF) feature weighting and radial basis function (RBF) neural networks. A subset of 32,715 college-student comments from Tsinghua Universitys Weibo-100k dataset serves as training samples, with cross-domain validation performed using the ChnSentiCorp benchmark. The approach optimises text feature sparsity via TF-IDF and utilises the nonlinear classification capability of RBF networks for opinion risk categorisation. Experimental results demonstrate an F1-score of 89.7% on the test set - marking a 6.2% improvement over conventional long short-term memory networks - while reducing warning response latency to 12 ms. This confirms high accuracy and real-time performance, providing a lightweight solution for monitoring campus ideological dynamics.
  Keywords: public opinion early warning; TF-IDF features; radial basis neural network; college students’ thought dynamics; social media analysis.
  DOI: 10.1504/IJICT.2026.10076066
   
  

• Open Access

  College students career planning for the development of low-carbon renewable energy economy ( Free Full-text Access )

  by Yunwei Dong 
  Abstract: This study examines the impact of low-carbon renewable energy economic development on college students' career planning. Findings reveal rapid industry growth but a significant talent shortage, with only 15% of students considering careers in this sector due to limited awareness. The paper proposes enhancing industry promotion, improving relevant knowledge and skills, expanding internships and employment channels, and calls for governmental and societal support to foster sustainable industry growth and talent cultivation.
  Keywords: low carbon economy; employment; renewable energy; market research; career planning; survey research.
  DOI: 10.1504/IJICT.2026.10076117
   
  

• Open Access

  The synergy of educational resource allocation and teacher motivation based on NSGA-II model ( Free Full-text Access )

  by Xueyao Wang, Jin Xu 
  Abstract: This article proposes a collaborative optimisation model for educational resource allocation and teacher incentive mechanism based on NSGA II. By simulating various allocation and incentive strategies, the model quantitatively analysed their interactions. The results indicate a significant synergistic effect: optimising coordination can improve student performance and teacher efficiency. Once the incentive intensity reaches the threshold, the teaching quality and teacher participation significantly improve, while the turnover rate decreases. Research has shown that combining appropriate resource allocation with incentive design can effectively improve educational outcomes. This method provides a scientific basis for resource allocation, offers a new perspective for incentive mechanism design, and has significant practical application value.
  Keywords: NSGA-II model; educational resource allocation; teacher motivation; multi-objective optimisation; quality of teaching.
  DOI: 10.1504/IJICT.2026.10076180
   
  

• Open Access

  Multimodal deep learning for evidence assessment with algorithmic bias analysis in criminal law ( Free Full-text Access )

  by Huishan Li 
  Abstract: This study integrates multimodal deep learning techniques for evidence assessment to investigate algorithmic fairness in the criminal justice system. The proposed approach predicts criminal charges and evaluates bias related to age and ethnicity by analysing demographic data, online crime reports, and historical records. Convolutional and recurrent neural networks with fairness-aware regularisation are employed to balance equity and predictive accuracy. While algorithmic crime prediction can assist judicial decision-making, it often faces criticism for bias, limited transparency, and lack of interpretability. The primary objective of this research is to predict charge severity while ensuring fairness and transparency. Prior studies have emphasised deep learning applications in fairness-aware algorithms and legal decision prediction, as well as potential racial bias in tools like COMPAS. Using extensive government statistics and crime narratives, ConvLSTM and Bi-LSTM models achieved superior performance, with macro-average F1 scores up to 0.86, while fairness regularisation reduced demographic disparities.
  Keywords: algorithmic crime prediction; deep learning models; bi-LSTM/RNN; ConvLSTM architecture; neural network classifiers; risk assessment instruments; RAIs; algorithmic fairness; bias in criminal justice.
  DOI: 10.1504/IJICT.2026.10076181
   
  

• Open Access

  Optimal allocation model network of human resources in energy enterprises based on NSGA-II ( Free Full-text Access )

  by Zhenhua Fan 
  Abstract: Against the backdrop of the dual carbon; goal and the intelligent transformation of energy systems, energy enterprises face core human resource allocation challenges, including 60% fluctuations in demand between peak and valley, multi-objective optimisation, and complex skill matching. Traditional static methods lead to a 40%-50% labour cost ratio and an over 35% skill mismatch rate. This study proposes a dynamic optimisation model with a forecasting-optimisation-real-time adjustment closed-loop framework, adopting ARIMA-GARCH (+-7% error) and IE-NSGA-II for four-objective optimisation. Empirical tests on a provincial power grid show the model reduces labour costs and carbon emissions by 17.3% and 10.5%, respectively, while improving efficiency and satisfaction by 21.8% and 18.6%, respectively.
  Keywords: Against the backdrop of the ‘dual carbon; goal and the intelligent transformation of energy systems; energy enterprises face core human resource allocation challenges; including 60% fluctuations in demand between peak and valley; multi-objective optimisation; and complex skill matching. Traditional static methods lead to a 40%–50% labour cost ratio and an over 35% skill mismatch rate. This study proposes a dynamic optimisation model with a ‘forecasting-optimisation-real-time adjustment’ closed-loop framework; adopting ARIMA-GARCH (±7% error) and IE-NSGA-II for four-objective optimisation. Empirical tests on a provincial power grid show the model reduces labour costs and carbon emissions by 17.3% and 10.5%; respectively; while improving efficiency and satisfaction by 21.8% and 18.6%; respectively.
  DOI: 10.1504/IJICT.2026.10076333
   
  

• Open Access

  AI-powered recommendation and task assignment mechanism for interactive vocational English teaching ( Free Full-text Access )

  by Yanxia Geng, Meilan Jin 
  Abstract: This study presents an AI-powered recommendation and task assignment mechanism designed to enhance interactive vocational English teaching. Making use of NLP, machine learning methods, and massive language models, the system personalises learning by analysing student proficiency, learning styles, and task performance. The proposed framework incorporates modules for content creation, personalised learning, and adaptive recommendations, supported by features such as passage and video wizards. Data collected from 500 students across rural and urban areas was processed to generate tailored learning paths, with performance evaluated using various regression models. Results indicate that the Huber Regress or achieved the highest predictive accuracy, enabling dynamic adjustments to learning tasks. The system demonstrated improved engagement and learning outcomes, particularly in contexts promoting learner-generated content and autonomy. These results demonstrate the promise of AI-powered platforms to provide practical, scalable language instruction.
  Keywords: artificial intelligence; recommendation system; task assignment; vocational English teaching; ML; personalised learning; educational technology; learner-generated context.
  DOI: 10.1504/IJICT.2026.10076064
   
  

• Open Access

  Construction of digital art knowledge graph based on deep recurrent neural network ( Free Full-text Access )

  by Huan Wang 
  Abstract: This study presents a method for constructing digital art knowledge graphs based on deep recurrent neural network (DRNN). A digital art knowledge graph is initially constructed by extracting visual features with ResNet50 and identifying textual entities via a CNN-BiLSTM-CRF model. Then, a DRDA model with bidirectional gated recurrent unit (GRU) and neighbour-aware attention is proposed for graph completion. Experiments on DBPedia50k and DBPedia500k show DRDA's superiority over three baselines. On DBPedia50k, DRDA improves head prediction MRR by up to 55% and achieves the lowest MR in tail prediction, though trailing slightly in Hits@10. On DBPedia500k, DRDA consistently outperforms baselines with MR reductions of 59-406 and MRR gains of 2%-19%. Further analysis identifies optimal depth and neighbour parameters, validating the model's scalability and its effectiveness in capturing complex semantic dependencies in large-scale multimodal art data.
  Keywords: digital art; knowledge graph; deep recurrent neural network; DRNN.
  DOI: 10.1504/IJICT.2026.10076063
   
  

• Open Access

  Real-time AI-regulated animation-user interaction system in virtual reality environments ( Free Full-text Access )

  by Jun Liu 
  Abstract: This study presents a real-time AI-regulated animation and user interaction system that leverages machine learning to enhance immersion in virtual reality (VR) environments. The framework integrates real-time simulation, CAD optimisation, and AI-driven animation to deliver responsive, realistic, and user-friendly interactions. Although challenges remain in resource utilisation and frame rate stability, experimental evaluations demonstrate high accuracy, responsiveness, and usability. The findings suggest that AI-governed VR systems hold significant potential in education, healthcare, and training for high-risk environments. As VR adoption expands across medicine, education, and the arts, the need for machines capable of dynamically controlling interactions and animations becomes critical for achieving presence and adaptability. Earlier rule-based and tele-operated approaches lacked realism and scalability, while newer AI-powered methods offer greater flexibility yet still face integration challenges. This system addresses these limitations by combining AI models with CAD-optimised animations, ensuring speed, precision, and usability in real-time.
  Keywords: augmented reality; AR; data mining algorithms; interaction with users; AI-regulated rendering; computer-aided design; CAD; optimisation; immersive reality; VR.
  DOI: 10.1504/IJICT.2026.10076062
   
  

• Open Access

  Virtual reality and actual technology in psychology: intermediate research and analysis ( Free Full-text Access )

  by Guiying Li 
  Abstract: This study examines the therapeutic, interactive, and immersive potential of virtual reality (VR) as a transformative tool in psychology and education. Research indicates that VR enhances engagement, reduces stress, and supports learning through analysis of user experience, physiological data, and feedback. The findings highlight its value for experiential learning and mental health rehabilitation. Within a short period, VR has emerged as a revolutionary technology across medicine, academia, and the arts, offering new opportunities for therapy, education, and user engagement. While prior studies confirm VR's ability to improve learning outcomes and engagement, they also note challenges in content creation, particularly for non-technical users. Employing a mixed-method approach, this study collected both quantitative (questionnaires, physiological measures) and qualitative data. Results revealed high satisfaction (average recommendation score: 8.31/10), with physiological markers - hyperventilation (M = 96.36%) and resting heart rate (M = 76.32 bpm) - demonstrating VR's capacity to relax and engage users.
  Keywords: virtual reality; actual technology; psychology; immersive learning; empathy; rehabilitation; stress detection; human-computer interaction; educational technology; cognitive engagement.
  DOI: 10.1504/IJICT.2026.10076061
   
  

• Open Access

  Spatial domain semantic collaborative recognition model for complex emotions in artistic images ( Free Full-text Access )

  by Jing Wang, Dali Zou 
  Abstract: Oil paintings, watercolours and digital art convey human emotions. Complex emotions when visual elements blend with semantic information. Existing methods have three flaws: over reliance on low-level visual features misjudges serene loneliness; treating emotions as discrete labels misses ambiguity; and poor genre adaptability. This study proposes the spatial domain semantic collaborative recognition model for art complex emotions, via a dual-branch framework: spatial branch uses multi-scale convolutional neural network for global features, and semantic branch adopts graph attention network for semantic links. A cross-branch attention mechanism tunes visual; a Gaussian mixture model-based module quantifies emotion distribution. Experiments on two self-built datasets and public ArtEmis show: vs. traditional convolutional neural network and single-semantic models, it boosts accuracy by 28.3%, cuts mean absolute error by 32.1%, and maintains over 89% cross-genre accuracy. This work bridges the semantic-visual-emotional gap, supporting intelligent art curation, emotional interaction design and art therapy.
  Keywords: artistic image; complex emotion recognition; spatial-semantic collaboration; graph attention network; Gaussian mixture model; style adaptability.
  DOI: 10.1504/IJICT.2026.10076065