International Journal of Information and Communication Technology (24 papers in press)

Regular Issues

• Open Access

  The prediction model of higher vocational students' classroom participation based on the fusion of deep learning and support vector machine ( Free Full-text Access )

  by Yixuan Qiang 
  Abstract: Student engagement in vocational classrooms is a critical metric for assessing teaching effectiveness and talent development. To address the limitations of conventional assessment methods, we propose a hybrid deep learning-support vector machine (SVM) model for predicting participation levels. The approach integrates convolutional neural networks (CNN) and long short-term memory (LSTM) networks to extract high-dimensional temporal features from classroom videos and behavioural logs. These features are combined with traditional statistical indicators and classified using SVM through a feature-level fusion strategy. Evaluated on simulated vocational classroom data, the fused model achieves 92.3% accuracy and an F1-score of 0.914, significantly outperforming standalone CNN-LSTM or SVM models. This model enables real-time, quantitative assessment of classroom engagement and supports timely teaching interventions.
  Keywords: classroom participation; deep learning; support vector machine; SVM; feature fusion; vocational education.
  DOI: 10.1504/IJICT.2026.10075999
   
  

• Open Access

  Generative music composition teaching system based on mobile interaction ( Free Full-text Access )

  by Yuanyuan Xie, Xiaole Zhu, Xing Hu 
  Abstract: Current music composition teaching systems face challenges such as limited resource quality and insufficient real-time interactivity. To address these issues, this paper proposes a generative music composition teaching system based on mobile interaction technology. The system first designs a music teaching resource generation module utilising multi-scale feature filtering, combined with a multi-discriminator structure to enhance the discriminative capability of generated samples. Building upon the generation of rich music teaching resources, this paper introduces an expandable network communication module and an interactive collaboration module supporting bidirectional collaborative control and user state management. Experimental results demonstrate that the designed system achieves a CPU utilisation rate of 43% and a single interaction response time of only 11.8ms. It not only generates high-quality music creation resources but also exhibits outstanding real-time interactive performance, holding significant value for advancing the widespread application of interactive mobile teaching.
  Keywords: music composition teaching; music generation; mobile interaction; feature selection; collaborative learning.
  DOI: 10.1504/IJICT.2026.10075963
   
  

• Open Access

  Temporal convolutional networks with language models for decoding music preferences in mental health profiling ( Free Full-text Access )

  by Junmei Bai 
  Abstract: Music preferences serve as crucial behavioural clues for decoding mental health states. Music provides a continuous and emotionally rich behavioural signal that is less influenced by social desirability biases compared to self-reported data, making it a robust indicator for mental health assessment. However, traditional analysis methods struggle to simultaneously account for the temporal dynamics of music listening and its rich semantic information, resulting in limited decoding efficacy. Previous studies attempted hybrid models but often faced overfitting or computational inefficiency, which motivated our design of a more integrated framework. To address this, we propose an innovative framework that integrates temporal convolutional networks with pre-trained language models to capture both the sequential patterns of music consumption and the emotional semantics of lyrics content. Our validation on a public dataset containing over 100,000 records demonstrates that this model achieves approximately 8.5% higher accuracy than single-modal benchmark methods in mental health state assessment tasks. It also effectively identifies specific musical features associated with depressive and anxious tendencies. This work provides a novel technical pathway for achieving non-invasive, dynamic mental health screening.
  Keywords: temporal convolutional networks; language models; music preferences; mental health; multimodal fusion.
  DOI: 10.1504/IJICT.2026.10075964
   
  

• Open Access

  Research on the identification and optimisation of traditional cultural symbols from the perspective of cross-cultural communication ( Free Full-text Access )

  by Anzhu Li 
  Abstract: This study looks at how to spot and use traditional cultural symbols in a way that helps people from different cultures talk to each other. The five-step plan is based on data gathering, communication studies, design theory, and semiotics. Steps are recognising symbols, analysing across cultures, making things better, and confirming. It is culturally rich and up-to-date because it is based on reviews from experts, feedback from the public, and how well it does in the market. We compared 40 grey relational analysis (GRA) and fuzzy comprehensive evaluation (FCE) models side by side and discovered that blue and white porcelain designs with lotus, peony, and plum flowers are the most attractive, culturally significant, and marketable. The results show how important it is for images to be clear, for societies to be open to change, and for people to be able to see things in different ways. The difference between the Chinese loong and the Thai naga shows how important it is to use symbols in world design in a way that makes sense all the time. This study gives us ideas for mixing old customs with new tools that are both useful and interesting.
  Keywords: cross-cultural communication; semiotics; cultural symbols; white and blue porcelain; symbol optimisation.
  DOI: 10.1504/IJICT.2026.10075965
   
  

• Open Access

  Application of an AI-driven visual aesthetic scoring system for style calibration in art works ( Free Full-text Access )

  by Feng Tan, Mei Wang 
  Abstract: This research explores AI-driven visual aesthetic scoring systems as tools for evaluating and refining artistic styles, with a particular focus on interior design and computational modelling. The study demonstrates how artificial intelligence can enhance artistic quality and align computer-generated imagery with human aesthetic preferences. By integrating compounded loss functions, curated datasets, and diffusion-based architectures, the model significantly improves visual appeal, stylistic consistency, and task performance. A composite loss-based AI framework was developed using a customised interior design dataset annotated with style tags, aesthetic ratings, and spatial attributes. The system, fine-tuned with user-defined parameters, produced results that were both visually appealing and contextually appropriate. Experimental outcomes revealed statistically robust improvements of 52.54% in portal engagement (Cohens d = 1.69, p < 0.001, 95% CI: [47.8%, 57.3%]) and 40.08% in agency engagement (d = 1.52, p < 0.001), validated through rigorous statistical testing including permutation tests, bootstrap resampling, and multiple comparison corrections. User studies further indicated that AI-selected or AI-generated images were preferred over other sources, receiving higher aesthetic ratings and engagement levels.
  Keywords: visual aesthetic scoring; style calibration; AI in art; diffusion models; aesthetic evaluation; artistic style transfer; generative design; computational aesthetics.
  DOI: 10.1504/IJICT.2026.10076004
   
  

• Open Access

  Innovative practice of AI-driven intelligent assessment system in university course teaching reform ( Free Full-text Access )

  by Minghui Sun, Jilai Liu 
  Abstract: This research explores the implementation of intelligent evaluation systems powered by artificial intelligence within the context of university teaching reform. By integrating convolutional neural networks with interactive internet of things - enabled systems, the study demonstrates significant improvements in student performance, grading efficiency, and learning outcomes. AI advancements are driving a major transformation in higher education, enabling efficient assessment and personalised learning opportunities. While previous research has examined AI in intelligent tutoring, adaptive learning, and automated assessments, comprehensive studies on its integration into higher education remain limited. Employing a mixed-method approach, this study collected data from 120 students before and after the intervention, using both quantitative tests and qualitative surveys to ensure thorough analysis. Results indicate that student satisfaction increased, grading time was reduced by over 40%, and test performance improved by 6%. The findings reveal that AI integration was positively received by both faculty and students.
  Keywords: artificial intelligence; AI; intelligent assessment; higher education reform; convolutional neural networks; CNNs; IoT–assisted systems; student performance.
  DOI: 10.1504/IJICT.2026.10076005
   
  

• Open Access

  Construction of mental health analysis model based on multi-modal feature learning and fusion network ( Free Full-text Access )

  by Sujing Li, Suya Liu, Maochun Wu 
  Abstract: This paper presents a mental health analysis model using a multi-modal feature learning and fusion network to improve assessment accuracy. It integrates data from text, images, and speech, processed with CNNs, RNNs, and LSTMs for feature extraction and fusion. Experimental results show the multi-modal model achieves 85% classification accuracy, outperforming single-modal models (75%). Analysis of feature weights indicates audio and visual modalities significantly influence emotional fluctuation (30%) and coping ability (40%), while physiological signals are crucial across all traits. The model enhances assessment comprehensiveness and offers effective support for early diagnosis and personalised intervention.
  Keywords: multimodal feature learning; deep learning; mental health analysis; feature fusion; privacy protection.
  DOI: 10.1504/IJICT.2026.10076006
   
  

• Open Access

  Development of an AI-assisted spoken language assessment system for Japanese language teaching ( Free Full-text Access )

  by Li Zhang 
  Abstract: This project develops an AI-assisted spoken language assessment system to enhance Japanese language instruction. By integrating voice recognition, pronunciation analysis, and fluency scoring, the system provides reliable evaluations comparable to those of professional ratters. Utilising multilingual datasets and data augmentation, it reduces language learning anxiety and improves recognition accuracy. Since language anxiety negatively affects second-language acquisition particularly in oral proficiency this study aims to support Japanese learners and increase their speaking confidence. While prior research has demonstrated the benefits of ICT tools and flipped classrooms for pronunciation self-monitoring, limited studies have applied AI for comprehensive oral evaluation. The proposed four-step methodology includes data collection, feature extraction, model development, and validation. Informed by sentiment analysis and multilingual corpora, the system achieved an accuracy of 97.3% using a two-stream LSTM model, while translation-based augmentation improved Japanese sentiment analysis accuracy by 6.58%.
  Keywords: AI-assisted assessment; spoken language evaluation; Japanese language teaching; speech recognition; natural language processing; NLP; automated scoring; pronunciation analysis.
  DOI: 10.1504/IJICT.2026.10076058
   
  

• Open Access

  Federated learning-enabled personalised delivery and student privacy protection in universities ( Free Full-text Access )

  by Yunxian Li 
  Abstract: This study presents a federated learning framework enhanced with entropy-adaptive differential privacy, blockchain consensus, and knowledge distillation to safeguard student data while improving personalised education. Traditional federated learning preserves privacy by training collaboratively without sharing raw data, yet faces challenges of heterogeneity, efficiency, and resilience against malicious clients. Existing solutions like homomorphic encryption and secure multiparty computation often incur high computational costs and limited adaptability. To address these limitations, the proposed framework employs blockchain-based role incentives to ensure fairness and verifiability, while entropy-adaptive differential privacy dynamically balances privacy and utility. Knowledge distillation further improves robustness and mitigates non-IID data distribution issues. Experiments on a Python programming course dataset with 2,452 students demonstrate superior accuracy, fairness, and resilience compared to conventional FedAvg. The method achieves up to 97% prediction accuracy with enhanced stability under adversarial conditions, offering a scalable and secure solution for personalised, privacy-preserving education.
  Keywords: federated learning; FL; personalised federated learning; PFL; student privacy protection; differential privacy; DP; homomorphic encryption; HE; blockchain-based federated learning; entropy-adaptive differential privacy; EADP; secure multi-party computation; SMPC.
  DOI: 10.1504/IJICT.2026.10076059
   
  

• Open Access

  Evaluation of teaching effectiveness in data analysis courses using a behavioural big data model ( Free Full-text Access )

  by Yixu Wang 
  Abstract: This study proposes a behavioural big data model for evaluating teaching effectiveness in data analysis courses across primary, secondary, and higher education levels. The framework integrates learning management system data, classroom engagement indicators, and student interaction behaviours to provide a comprehensive understanding of how teaching strategies influence learning outcomes. At the primary level, the model captures early learning patterns such as task attention, problem-solving attempts, and basic data reasoning through gamified digital activities. For secondary students, behavioural indicators including learning persistence, collaboration, and response accuracy are used to assess the development of analytical thinking and computational skills. Machine learning and statistical techniques, such as clustering, regression, and correlation analysis, identify patterns linking teaching approaches with student performance and motivation. With a predictive accuracy of 89%, the model demonstrates strong adaptability across age groups. Findings show that interactive, feedback-rich, and project-based learning environments significantly enhance students comprehension and retention of data analysis concepts.
  Keywords: data analysis education; behavioural big data; learning analytics; teaching effectiveness; student performance evaluation; big data in education; data-driven pedagogy.
  DOI: 10.1504/IJICT.2026.10076060
   
  

• 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 VRs 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 VRs 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

  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 teleoperated 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

  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 DRDAs 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 models 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

  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

  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; 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, 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 art emis show: vs traditional convolutional neural network and single-semantic models, it boosts accuracy by 28.3%, cuts mean absolute error by 32.1%, 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
   
  

• 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

  Forecasting trend of agricultural talents flow by spatio-temporal graph neural network and LightGBM ( Free Full-text Access )

  by Jie Liu 
  Abstract: Current agricultural talent flow prediction mainly uses single models (e.g., linear regression, ARIMA, LSTM), which fail to capture non-Euclidean spatial-temporal relationships and automatically extract complex spatio-temporal interactions, limiting accuracy and interpretability. This paper proposes a hybrid framework integrating a spatio-temporal graph neural network (STGNN) and LightGBM. Using 2010-2020 data from 17 cities in Henan Province, a spatio-temporal graph is built with city nodes and geographic-threshold edges. STGNN combines graph convolution and temporal convolution (TCN) to automatically learn spatio-temporal features, while LightGBM regresses lag and socio-economic indicators for interpretability. Benchmark comparisons with ARIMA, LSTM, and LightGBM, plus ablation and sensitivity tests, confirm the hybrid model's superiority. It reduces error by 10%-14% versus standalone STGNN/LightGBM, achieving under 12.3% overall error, with significantly improved accuracy and stability.
  Keywords: agricultural talent flow; spatio-temporal graph neural network; STGNN; LightGBM; hybrid prediction.
  DOI: 10.1504/IJICT.2026.10075962
   
  

• Open Access

  Prediction of uncertain passenger flow in scenic spots by fusing multi-source data and integrated learning ( Free Full-text Access )

  by Jingwen Xu, Qingshan Xiao, Shuo Xiong 
  Abstract: Accurate scenic spot traffic prediction is of great significance for the optimal allocation of tourism resources and safety management. Aiming at the shortcomings of traditional methods in coping with data multi-source and prediction uncertainty, this study proposes an uncertainty prediction framework that integrates multi-source data and integrated learning. By integrating heterogeneous data from multiple sources, such as historical passenger flow, meteorology, web search and spatial features, a heterogeneous integrated model based on random forest, XGBoost and long short-term memory (LSTM) is constructed, and quantification of uncertainty is realised by combining quantile regression and conformal prediction method. Experiments on public datasets show that this method reduces the mean square error (MSE) by 30%, the mean absolute percentage error (mean absolute percentage error) by 25%, and the prediction interval coverage reaches 95.3%, which provides reliable decision support for the intelligent management of scenic spots.
  Keywords: passenger flow prediction; multi-source data fusion; integrated learning; uncertainty quantification; tourist attractions.
  DOI: 10.1504/IJICT.2026.10075961
   
  

• Open Access

  Natural language processing for automatic error detection in Chinese language learning ( Free Full-text Access )

  by Zhengxin Li, Rongzhen Wu 
  Abstract: With the rapid development of global Chinese language education, the demand for efficient and accurate automated teaching assistance tools is growing. Traditional manual grading methods are often time-consuming and yield inconsistent results, highlighting the necessity for intelligent technological solutions. This paper explores the application of natural language processing techniques in automatic error detection for Chinese as a second language. It proposes a method based on pre-trained language models and evaluates it using a publicly available corpus of Chinese learner compositions. Experimental results demonstrate the strong performance of the proposed method in identifying grammatical and lexical errors, achieving detection accuracy exceeding 80% for major error categories. This represents a significant improvement over baseline systems (over 25% increase). This technology shows great potential as an efficient teaching support tool, enabling more effective and consistent feedback mechanisms within intelligent educational environments.
  Keywords: natural language processing; NLP; Chinese language teaching; automatic bias detection; applicability analysis.
  DOI: 10.1504/IJICT.2026.10075960
   
  

• Open Access

  Analysis of tourist emotions and behaviour patterns using deep learning ( Free Full-text Access )

  by Limin Wang 
  Abstract: This study explores the emotions and behavioural patterns of travellers using advanced deep learning techniques. A hybrid model combining CNN and LSTM networks was developed to analyse Twitter data related to travel in Thailand, enabling the identification of key emotional states such as joy, surprise, fear and melancholy. The proposed approach provides valuable insights for recommendation systems and tourism management, as tourists' emotions significantly influence travel decisions and satisfaction levels. By analysing emotional tendencies, tourism services can enhance the overall visitor experience. While previous research has largely relied on machine learning and lexicon-based methods for textual emotion detection, recent advancements in deep learning have demonstrated superior predictive accuracy for sentiment and behavioural analysis. In this study, CNN-LSTM models, complemented by feature extraction techniques using DenseNet and AlexNet, were employed. The hybrid model achieved 91% accuracy, surpassing conventional methods, with joy and surprise being the most accurately classified positive emotions.
  Keywords: tourist emotions; behaviour patterns; emotion recognition; tourist experience analysis; deep learning; behavioural analytics; artificial intelligence in tourism.
  DOI: 10.1504/IJICT.2026.10075959
   
  

• Open Access

  Personalised learning path generation mechanism based on RL and knowledge graph ( Free Full-text Access )

  by Zhao Liu, Yanan Shang 
  Abstract: This article proposes a personalised learning path generation mechanism that combines reinforcement learning and knowledge graphs. It constructs a knowledge graph that includes knowledge points, student status, and historical behaviour. Using this image and learning trajectory, it constructed a student model. Then, the RL algorithm optimises the learning path based on real-time feedback. An experiment targeting 300 college students compared the proposed method with traditional methods. The results showed that reinforcement learning based methods improved learning outcomes by 12.5%, increased learning satisfaction by 23.7%, accelerated knowledge acquisition by 15%, and shortened average learning time by 8%. These findings confirm the effectiveness of this mechanism in improving learning outcomes and meeting individual student needs.
  Keywords: personalised learning; reinforcement learning; knowledge graph; learning pathways; data analysis.
  DOI: 10.1504/IJICT.2026.10075958