International Journal of Information and Communication Technology (12 papers in press)

Regular Issues

• Open Access

  Correlation analysis between local government debt and economic growth combined with PSTR model ( Free Full-text Access )

  by Lian Pan 
  Abstract: This paper innovatively combines PSTR model with federal learning data enhancement algorithm, further improves the data processing efficiency of PSTR model through computer data processing method, enriches and improves the theoretical research on industrial structure and local government debt (LGD). This paper verifies through empirical research that the correlation analysis method between LGD and economic growth combined with PSTR model has certain effects. Based on the research conclusions, we can put forward corresponding policy suggestions on how local governments should maintain scientific debt scale to give full play to the intermediary effect of industrial structure and promote economic development.
  Keywords: PSTR model; local government debt; LGD; economic growth; relevance.
  
  

• Open Access

  Prediction of talent demand and job matching based on knowledge graph and attention mechanisms ( Free Full-text Access )

  by Xiaoli Mei 
  Abstract: The real-time job data from recruitment platforms, reflecting the demands of enterprises for job seekers, can provide data support for the development of university training policies. This study proposes a knowledge representation model based on the self-attention mechanism for talent demand and job matching prediction method, where neural networks aggregate neighbourhood information to generate better node representations so that nodes can learn their local neighbourhood information and the whole graph structure, and at the same time, using the self-attention mechanism, further extract node features containing rich neighbourhood information, and use deep interactions between nodes to calculate the central entity around the attention coefficients of neighbours to improve the accuracy of prediction. The experimental results show that the model prediction is highly fault-tolerant, short time-consuming, and can be widely used in the practical application of matching talent demand and university talent cultivation.
  Keywords: knowledge graph; attention mechanism; talent demand; job matching.
  
  

• Open Access

  Single-stage landmark retrieval with texture feature fusion ( Free Full-text Access )

  by Kun Tong, GuoXin Tan 
  Abstract: Existing landmark retrieval models typically fuse global and local feature descriptors of target images to generate feature vectors for landmark retrieval. However, these models often exhibit poor resilience to complex viewpoints, occlusions, and lighting conditions. Moreover, the fused feature descriptors still contain substantial redundant information, leading to decreased retrieval accuracy. To address these issues, this paper proposes a novel single-stage image retrieval model enhanced by texture augmentation. The model incorporates a texture enhancement module that leverages texture feature encoding to reconstruct the original feature maps, amplifying the influence of texture features in deep feature vectors across different scales. This approach ensures robust feature representation under extreme angles, occlusions, or varying lighting conditions. To mitigate the problem of redundant features, the model introduces an innovative feature fusion module. This module optimizes local features from multi-scale feature descriptors using a mapping fusion technique, eliminating redundant information and generating more compact and discriminative feature descriptors. Extensive experiments demonstrate that the proposed model achieves significant improvements in retrieval performance compared to state-of-the-art image retrieval models, while maintaining acceptable retrieval times.
  Keywords: landmark retrieval; feature fusion; multi-scale fusion; feature enhancement.
  
  

• Open Access

  Stereo vision and deep learning-based movement correction for sports dance ( Free Full-text Access )

  by Wumei Jiang, Hui Lan 
  Abstract: In sports dance training, dancers’ wrong movements are unavoidable, and if they are not corrected in time, it will not only reduce the effect of dance expression, but also directly related to the improvement of sports performance. Therefore, we suggests a correction method for sports dance movements based on stereo vision and deep learning. Firstly, a binocular stereo imaging model is established by using the principle of triangle similarity. Secondly, 3D CNN is used to extract spatio-temporal features from the preprocessed images, and the early attention mechanism is introduced to adaptively enhance the key features that are beneficial to early action prediction. Finally, the important features are used to model the action boundaries by estimating the relative probability distribution of the action boundaries to obtain the recognition results. Simulation experiments show that the accuracy and peak signal-to-noise ratio are 91.17% and 20.45 dB, respectively.
  Keywords: stereo vision; deep learning; motion correction; 3D CNN; attention mechanism.
  
  

• Open Access

  A meta-heuristic optimisation algorithm based method for scheduling edge computing resources ( Free Full-text Access )

  by Yujie Li, Yaoyao Xu, Fangfang Cao, Xiang He 
  Abstract: Edge computing provides a viable solution to the lack of computing power in smart mobile devices (SMDs) and has received much attention in the industry. However, transferring part of the computation tasks from SMDs to edge servers brings additional transmission energy and server computation energy. To reduce energy consumption, this article suggests an edge calculating resource scheduling approach relied on meta-heuristic improvement algorithm. Firstly, the resource scheduling system model is constructed, and the SMD selects the most suitable edge server (ES) to help itself to complete the computational tasks according to the computational resources of the ES. Then the total energy consumption objective function is suggested, and an enhanced particle swarm optimisation (EPSO) algorithm is used to address this objective function. The experimental outcome indicates that when the number of SMDs is 10, the energy consumption value of the suggested method is 9.25W, which is reduced by 10%-55% compared to the other four methods.
  Keywords: resource scheduling; metaheuristic optimisation algorithm; particle swarm optimisation algorithm; power law distribution function; genetic algorithm; smart mobile devices; SMDs; enhanced particle swarm optimisation; EPSO.
  
  

• Open Access

  The application of big data analysis in logistics supply chain optimisation ( Free Full-text Access )

  by Chunsheng Liu 
  Abstract: This article studies the application of big data in logistics supply chain optimization, combining modern data processing technology with intelligent optimization algorithms, aiming to improve supply chain efficiency and reduce logistics costs. The K-Means clustering algorithm is used to partition the delivery area and customer demand, in order to optimize the allocation of logistics resources and warehouse layout, and reduce the redundancy of transportation paths. Next, based on the clustering results, the ant colony algorithm is utilized to address the optimization of vehicle routing problem (VRP), finding the shortest path between multiple delivery points to minimize transportation time and cost. This article utilizes the big data analysis platform Hadoop for data storage and processing, ensuring the efficient operation of algorithms on large-scale data. The results show that the supply chain optimization strategy combining big data analysis, K-Means clustering, and ant colony optimization can improve delivery efficiency and reduce operating costs.
  Keywords: big data; ant colony algorithm; k-means; logistics chain.
  
  

• Open Access

  Mental toughness prediction model of college students based on optimal elastic network regression ( Free Full-text Access )

  by Fulian Liu 
  Abstract: Predicting the level of mental toughness can help colleges and universities better understand the psychological condition of college students. This paper designs a prediction model of college students mental toughness based on optimised elastic network regression (ENR) to address the redundant features as well as the overfitting problems of existing studies. Firstly, the ENR is optimised using Bayesian optimisation algorithm (BOENR). Secondly, the important influencing factors are extracted to the maximum extent by using the partial least squares method. Then, linear discriminant analysis (LDA) is used for feature screening of key influencing factors, Pearsons correlation coefficient is used to measure the redundancy relationship among features, and finally, BOENR estimation of regression coefficients is computed based on each feature sample separately. The experimental outcome indicates that the MSE and MAE of the designed model are reduced by 0.03950.2264 compared with the other five models.
  Keywords: mental toughness prediction; elastic network regression; ENR; Bayesian optimisation; partial least square; linear discriminant analysis; LDA.
  
  

• Open Access

  Analysing trends in trading behaviour in financial markets using deep learning algorithms ( Free Full-text Access )

  by Yiming Tan 
  Abstract: The complexity of financial markets demands analytical methods that capture non-linear and time-varying data characteristics. Traditional methods often fall short, prompting the use of deep learning, particularly LSTM, for its time series processing prowess. However, LSTMs struggle with long sequences due to vanishing or exploding gradients, leading to the loss of early data significance. To address this, we introduce the LSTM-AT model, which integrates LSTM with an attention mechanism to enhance its focus on key data aspects. Our model, trained on historical financial data, outperforms traditional methods in predicting market trends. Despite its high accuracy, the model faces challenges like overfitting and data labelling requirements. Future work will focus on improving interpretability and exploring its application across various financial markets.
  Keywords: LSTM-AT; financial markets; trading behaviour; trend prediction.
  
  

• Open Access

  Prediction method of battery remaining life based on CEEMDAN-RF-SED-LSTM neural network ( Free Full-text Access )

  by Hanping Pan, Wenjian Feng, Jiaqi Ge, Jijun Zhang 
  Abstract: In order to ensure the reliability and safety of electric vehicles, a CEEMDAN-RF-SED-LSTM method for lithium ion power battery system is proposed. Taking the time interval of equal voltage charging as the indirect health factor, taking into account the influence of external interference and capacity regeneration phenomenon, the degradation trend of battery is obtained by variational mode decomposition (VMD). The improved recurrent neural network model short and long time series (LSTM) is used to obtain the residual life prediction. Finally, the established model is compared with FNN, CNN, LSTM and other neural networks, which gives full play to the characteristics of SCN such as strong autonomy, fast convergence speed and low network cost. The performance of this method is tested with NASA dataset as the research object. The experimental results show that CEEMDAN-RF SED-LSTM model performs well in predicting RUL of batteries, and the prediction results have lower errors than that of a single model.
  Keywords: lithium ion battery; life prediction; configure the network randomly; incremental learning.
  
  

• Open Access

  Dynamic performance of a delayed-onset disease model including death and recovery ( Free Full-text Access )

  by Quanhao Song 
  Abstract: Due to the complex and constantly evolving process of infectious disease transmission, we have studied a class of delayed SIR models incorporating both death and recovery to control the spread of diseases. We creatively use bifurcation theories to determine the critical delay 0 for Hopf bifurcation to comprehend the impact of time delay on the system. The analysis of periodic solutions and bifurcation directions, based on central manifold and normal form theory, offers insights into the systems dynamics. Simulations utilising time series charts and trajectory diagrams aid in comprehending the impact of hysteresis parameters. Additionally, data fitting is performed to verify the proposed model by contrasting it with actual data. The research demonstrates that the implementation of consistent preventive and hygienic measures can significantly reduce the severity of disease exacerbation.
  Keywords: SIR model; basic reproductive number; time delay; Hopf bifurcation; centre manifold theorem.
  
  
• Interactive decision support system with machine intelligence for augmentative communication  
  by Ruiwei Chen, C.B. Sivaparthipan 
  Abstract: Many augmentative communication technologies help physically challenged people to communicate with others in the present world. Augmentative communication system integrates components that include symbols, strategies, and aids that enhance communication abilities. Augmentative communication technology’s significant challenge for physically challenged people is the lack of speech expression and depression. Interactive decision support system integrated machine intelligence framework (DSS-MIF) supports augmentative communication proposed to express the physically challenged expression and depression. DSS has multiple sensors, which monitor the heartbeat rate, vocal cord vibration, body temperature, and muscle contraction. The related data are calibrated using MIF, in which the expression of the person is recognised. Based on the DSS-MIF output, physically challenged people could express themselves to others using the augmentative communication system. The experimental analysis shows that the proposed DDS-MIF for augmentative communication improves performance rate to 98.66% and shows physically challenged people’s expression effectively.
  Keywords: augmentative communication; machine intelligence; decision support; multiple sensors.
  DOI: 10.1504/IJICT.2023.10056637
   
  
• A new fast DBSCAN using dual-space analysis and colour integral volume for document image segmentation  
  by Zakia Kezzoula, Djamel Gaceb 
  Abstract: The segmentation of the colour document images is an essential step allowing facilitating and improving the stages of characterisation and interpretation of the information contained in these documents. Recent systems of automatic processing and recognition of document images, which use separation of colouremric layers, are more efficient compared to conventional systems, only based on binary or grey levels images. This task requires non-supervised pixel segmentation or clustering techniques to separate the document image to a variable and unknown number of colour layers. The methods based on density are widely used in this context at pixel level, such as the DBSCAN method and its different variants, very robust to the noise and more adapted to the degradations present on document images, but who suffer from a great complexity. In this context, we propose a new faster DBSCAN variant using the volume integral in colourimitric space for the first time to significantly reduce calculation time. The combination of the two spaces, Cartesian and colorimetric has also accelerated the method and improved its performance on document images with different challenges. The results obtained show the effectiveness of the proposed approach, which was marked by significant improvement in the quality of segmentation and reduction in computed time.
  Keywords: clustering; DBSCAN; region growing; document image segmentation; fast I2SDBSCAN; 3D colour histogram; integral volume.
  DOI: 10.1504/IJICT.2024.10065387