International Journal of Business Intelligence and Data Mining (8 papers in press)

Special Issue on: Exploring AI Methods and Applications for Data Mining Part Two

• Open Access

  DNN and BiGRU-based hierarchical attention network for intrusion detection ( Free Full-text Access )

  by Hui Yan, Hupeng Liu, Ping Yu, Xiaoqing Xu, Mingxin Li, Yunxin Long, Hanlin CHEN, Qi Wang, Duo Long 
  Abstract: To tackle the rising sophistication of cyberattacks and the critical need to enhance intrusion detection capabilities, this study presents a hybrid architecture combining a DNN with an attention mechanism and a BiGRU with an attention mechanism, BiGRU with attention, and an MLP classifier. The model integrates dual DNN-Attention and BiGRU-Attention submodules to capture high-dimensional static features and temporal dependencies, followed by feature fusion and classification via MLP. Evaluated on NSL-KDD and UNSW-NB15 datasets, the model achieves validation accuracies of 99.48% and 98.43%, respectively, with stable convergence and low loss. Comparative results demonstrate superior performance in accuracy, robustness, and generalization, confirming its effectiveness for practical cybersecurity applications.
  Keywords: intrusion detection; deep learning; bidirectional gated recurrent unit; attention mechanism; network security.
  DOI: 10.1504/IJBIDM.2026.10079023
   
  

• Open Access

  Rapid planning of multimodal transport path based on improved VSRB-RRT algorithm ( Free Full-text Access )

  by Fang Wang, Chunzheng Zhao, Xiaoya Huang, Yali Liang 
  Abstract: To address the problems of insufficient network coverage, low on-time delivery rate, and long planning scheme generation time in traditional methods, a rapid planning method of multimodal transport path based on improved VSRB-RRT algorithm is proposed. Multi modal transportation network topology model, and GA-PSO hybrid algorithm for hub location optimisation, to build an efficient network foundation for path search; Based on the results of hub site selection, an improved VSRB-RRT algorithm was designed and implemented. The algorithm improves exploration efficiency through variable sampling areas and bidirectional search mechanism, adapts to complex environments with dynamic step size adjustment, and optimises path quality through node pruning and B-spline curve smoothing, achieving fast and reliable multimodal transportation path generation. Experimental results show that the proposed method has a multimodal transportation network coverage rate of up to 97.12%, a peak on-time delivery rate of 98.67%, and a minimum scheme generation time of only 3.61s.
  Keywords: improved VSRB-RRT algorithm; multimodal transport path; rapid planning; network topology; GA-PSO hybrid algorithm.
  DOI: 10.1504/IJBIDM.2026.10079262
   
  

• Open Access

  A resource allocation method for digital online teaching platform based on classification mining ( Free Full-text Access )

  by Weiya Xu, Xi Lin 
  Abstract: In the process of deepening the evolution of digital education environment, online teaching platforms often suffer from problems such as the generalisation of heterogeneous resources, dynamic demand changes, and imbalanced system architecture. Therefore, a resource allocation method for digital online teaching platforms based on classification mining is proposed. Firstly, based on the improved particle swarm optimisation algorithm, the twin support vector machine is optimised for resource classification. Secondly, the matching degree is quantified by the load difference and imbalance degree of physical nodes, and the virtual machine resource demand is constrained within the physical machine capacity. Finally, by combining real-time resource consumption and node performance indicators, a configuration function and adaptation factor are constructed to achieve balanced and optimised allocation of teaching resources. The test results show that the method maintains a stable resource allocation balance of over 90%, and the resource allocation response time is always below 1.5 seconds.
  Keywords: classification mining; digitisation; online teaching platform; resource allocation; twin support vector machine; TWSVM; particle swarm optimisation; PSO.
  DOI: 10.1504/IJBIDM.2026.10079431
   
  

• Open Access

  A mathematical model of attribute-based encryption for mining clusters in big data ( Free Full-text Access )

  by Yumei Zhao, Huaxi Chen 
  Abstract: Traditional encryption methods are difficult to meet the fine-grained requirements of multi-user collaboration and dynamic access in big data scenarios, and there are still shortcomings in terms of recall rate, attribute hiding success rate, and encryption efficiency. Therefore, this article proposes a mathematical modeling method for encrypting big data attributes based on clustering mining. The structured dynamic fusion deep graph clustering framework enables effective big data cluster analysis with comprehensive result processing and completion. A bilinear group-based cryptographic model incorporating attribute policies is constructed, encompassing fundamental procedures including system setup, key derivation, index encryption, and trapdoor computation, thereby establishing mathematical representation for attribute-based big data encryption. Experimental evaluations demonstrate the approach achieves 97.85% average recall, 98.7% successful attribute concealment rate, with merely 75.44ms average encryption latency.
  Keywords: cluster mining; big data attribute encryption; mathematical modelling; deep map clustering mining network; bilinear group.
  DOI: 10.1504/IJBIDM.2026.10079433
   
  

• Open Access

  Evaluation and optimisation method for graphic advertising design effectiveness based on binocular vision ( Free Full-text Access )

  by Jiaqi Li, Fang Wang, Huaxi Chen 
  Abstract: This article proposes a method for evaluating and optimising the design effect of flat advertisements based on binocular vision. The core innovation lies in simulating the binocular fusion mechanism of the human eye. By integrating Gabor energy response, saliency map, and disparity matrix, a comprehensive quality evaluation model is constructed, and targeted optimisation is driven based on this model. Generate intermediate views during the evaluation phase, extract global and local binocular visual features, combine monocular visual features, and quantify the design effect; during the optimisation phase, based on the evaluation results, visual consistency and stereoscopic perception are improved through image segmentation, colour space conversion, and edge roughness analysis. The experimental results show that this method maintains a stable accuracy of over 90% in visual attractiveness evaluation, with a minimum of 0.79 and a maximum of 0.87 after optimising the subject saliency, significantly better than the comparative methods.
  Keywords: binocular vision; print advertising design; effectiveness evaluation; optimisation method; visual perception.
  DOI: 10.1504/IJBIDM.2026.10079851
   
  

• Open Access

  Design of lightweight human pose estimation network for rehabilitation training ( Free Full-text Access )

  by Qingyun Yang, Gaigai Zhang 
  Abstract: Given the resource constraints of terminal devices, existing pose estimation networks, characterised by high computational complexity, struggle to satisfy the simultaneous demands of real-time performance and accuracy in rehabilitation training. Therefore, this paper proposes a lightweight human pose estimation network specifically designed for rehabilitation training. Firstly, laser triangulation is used to obtain the target image, and precise mapping from 3D to 2D space is achieved through coordinate transformation. Secondly, design an RMPE tiny lightweight network, introduce G-Bottleneck module to compress parameter quantity, and integrate Sa-ECA attention mechanism to enhance feature interaction. Finally, the Huber loss function is used to optimise training and improve convergence speed and robustness. The experimental results show that the method proposed in this paper maintains an overall accuracy of over 96% in human pose estimation testing, with most samples approaching or exceeding 98%, consistently maintains between 41.1 FPS and 44.4 FPS in ten inference speed tests.
  Keywords: rehabilitation training; lightweight network; human pose estimation; attention mechanism.
  DOI: 10.1504/IJBIDM.2026.10080082
   
  

• Open Access

  Study on high jump athlete error action recognition based on dual stream CNN BiLSTM ( Free Full-text Access )

  by Huifen Jia 
  Abstract: A high jump athlete error recognition method based on dual stream CNN BiLSTM is proposed to address the issues of rapid changes in high jump movements and strong temporal dependencies in multiple stages. Firstly, collect wrist, ankle, and waist movement data, and use an adaptive filtering algorithm driven by spectral kurtosis extremum for pre-processing. Secondly, a dual stream heterogeneous input architecture that integrates time-domain and frequency-domain features is constructed to process sensor timing signals and time-frequency spectra separately. Finally, CNN is used to extract spatial local features, and a bidirectional long short-term memory network (BiLSTM) is introduced to model the entire process of running, jumping, pole crossing, and landing in a long-range time series, achieving accurate recognition of erroneous actions. The experimental results show that this method outperforms the comparative methods in three indicators: top-1 accuracy (>96.8%), Cohens Kappa coefficient (>0.941), and average inference time (≤8.3 ms).
  Keywords: CNN BiLSTM model; high jumper; wrong action; image recognition.
  DOI: 10.1504/IJBIDM.2026.10080248
   
  

Special Issue on: Knowledge Discovery from Big Data to Spur Social Development Part 3

• A lightweight photovoltaic fault diagnosis model based on infrared images and texture features  
  by Zhuohang Wei , Xiaoyang Lu 
  Abstract: Fault diagnosis is crucial for the stability and safety of photovoltaic (PV) arrays' power generation. However, infrared image-based PV fault diagnosis faces challenges such as feature sparsity, scale differences, and sample imbalance. Thus, this paper proposes a lightweight PV fault diagnosis model based on infrared images to identify different fault types and severities. Firstly, an edge intensity embedding module is applied PV fault detection to enhance distinguishing faults of varying severities. Secondly, the proposed method adopts an improved MicroNet as the backbone to efficiently capture global thermal distribution features via sparse connection and dynamic cross-channel fusion. Subsequently, it combines PANet and multi-detection heads to improve detection robustness at different shooting distances. Finally, it uses a weighted loss function to alleviate sample imbalance. Experimental results show that with only 1.3 GFLOPs, the model achieves a mAP of 91.8%, outperforming other mainstream object detection algorithms.
  Keywords: PV fault detection; thermal imaging; MicroNet.
  DOI: 10.1504/IJBIDM.2027.10080247