Forthcoming and Online First Articles

International Journal of Reasoning-based Intelligent Systems

International Journal of Reasoning-based Intelligent Systems (IJRIS)

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 Articles will not be published until the final proofs are validated by their authors.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 published online here, before they appear in a journal issue. 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.

Open AccessArticles 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 Reasoning-based Intelligent Systems (86 papers in press)

Regular Issues

  • Image reconstruction of traditional Chinese painting works based on depth learning   Order a copy of this article
    by Xiyang Li, Wenzhen Ku 
    Abstract: Aiming at the problems of large sparse decomposition error of pixel signal, low reconstruction accuracy and slow reconstruction speed in traditional Chinese painting image reconstruction, a method of traditional Chinese painting image reconstruction based on depth learning is proposed. The image signal of traditional Chinese painting works is decomposed sparsely by the dictionary, the traditional Chinese painting image is weighted according to the Gaussian filter, the impurities in the signal are removed by the bilateral filter method, and the edge of the traditional Chinese painting image is corrected by the colour component method, so as to build the depth learning model of the traditional Chinese painting image reconstruction and achieve accurate and efficient reconstruction. The results show that this method can reduce the sparse decomposition error of pixel signal, less than 0.15%, the reconstruction accuracy reaches 97%, and the reconstruction time is shortened, the maximum time is only 1.5 s, indicating that this method is feasible.
    Keywords: deep learning; Chinese painting works; image reconstruction; Gaussian filtering; colour component; residual network.
    DOI: 10.1504/IJRIS.2023.10056191
     
  • An Innovative Educational Resource Allocation Method Based on NSGA-II Algorithm   Order a copy of this article
    by Jing Chen 
    Abstract: Because the traditional allocation method of innovative education resources has some problems, such as the closeness of innovative education indicators, the utilisation efficiency of innovative education resources and the low efficiency of innovative education resources allocation, proposes an innovative educational resource allocation method based on NSGA-II algorithm. Firstly, the index system of innovative education resource allocation is established, and the weight of the index is calculated by using the analytic hierarchy process. Secondly, with the goal of optimal closeness, maximum resource utilisation efficiency and fastest resource allocation efficiency, a multi-objective function of innovative education resource allocation is established. Finally, NSGA-II algorithm is used to solve the objective function, and the obtained solution is the optimal strategy. The experiment results show that the proposed method has a high closeness of innovative education indicators and a high efficiency of innovative education resource utilisation, which improves the efficiency of innovative education resource allocation.
    Keywords: NSGA-II algorithm; Innovation education; resource allocation; index closeness; multi-objective function.
    DOI: 10.1504/IJRIS.2023.10056192
     
  • The Perception Method of Sensitive Ideological and Political Teaching Resource Leakage Based on Stain Tracking   Order a copy of this article
    by Lina Zhao 
    Abstract: In order to solve the problems of large detection error, low data sensitivity and long perception time in sensitive ideological and political teaching resource leakage perception, a new sensitive ideological and political teaching resource leakage perception method based on stain tracking is designed. By building a collection model, the data of resources are collected, and the sensitivity of the collected data is determined with the help of cluster division method; The K-anonymity model is constructed to assume that the data vulnerability characteristics are determined, and the inconsistent and consistent eigenvalues are extracted. Initialise and mark data taint, set tracking rules and taint memory pool, build perception model with hash perception algorithm, calculate confidence, and implement method design. The results show that the leakage sensing error of this method is less than 2%, the data sensitivity is more than 90%, and the sensing time is only 9.71 s, which effectively reduces the sensing error and the sensing time, and improves the data sensitivity.
    Keywords: stain tracking; sensitive ideological and political education resources; leakage perception; cluster division; K-anonymous model.
    DOI: 10.1504/IJRIS.2023.10056193
     
  • Emotion recognition of digital art image based on weighted fusion strategy   Order a copy of this article
    by Xin Meng, Huili Tan, Xinyue Zhang 
    Abstract: In order to overcome the problems of low recognition accuracy and low recognition efficiency of traditional image emotion recognition methods, this paper proposes a digital art image emotion recognition method based on weighted fusion strategy. First, image emotional tags are designed, and image samples are selected using information entropy. Secondly, Gaussian fuzzy is used to reduce image noise and extract image emotional features. Then, the weighted fusion strategy is used to construct a weighted matrix to determine the similarity between feature classes; Finally, SVM classifier is constructed to classify image emotion features, emotion recognition function is designed according to weighted fusion strategy, and emotion recognition result is solved according to maximum rule. The results show that the recognition time of this method is less than 30 s, and the recognition accuracy can reach 99.0%, which shows that this method can improve the effect of emotion recognition.
    Keywords: weighted fusion; emotion recognition; Laplace operator; CLAHE algorithm; maximum rule.
    DOI: 10.1504/IJRIS.2023.10056194
     
  • An Online and offline hybrid teaching resource sharing method based on cloud computing   Order a copy of this article
    by Huishuang Qi 
    Abstract: In order to solve the problems of low resource request success rate, low data sharing ability and low sharing efficiency in existing resource sharing methods, this paper proposes an online and offline hybrid teaching resource sharing method based on cloud computing. Build a three-tier online and offline hybrid teaching resource sharing cloud platform, use Tucker decomposition algorithm to fuse the extracted features of teaching resources, and output them through the fuzzy neural network model; Build a teaching resource sharing model based on layered Agent technology to realise online and offline mixed teaching resource sharing. The comparative experiment verifies that the resource request success rate of the design method is always more than 80%, the maximum sharing of 98% of the total data can be achieved, and the sharing time is always less than 1 s, which proves that the design method in this paper has a high sharing efficiency.
    Keywords: cloud computing; online and offline; mixed teaching; resource sharing; tucker decomposition algorithm; fuzzy neural network.
    DOI: 10.1504/IJRIS.2023.10056195
     
  • Multimodal music emotion recognition method based on multi-source data fusion   Order a copy of this article
    by Bin Liu 
    Abstract: Aiming at the problems of low recognition accuracy and long recognition time in traditional multimodal music emotion recognition methods, a multimodal music emotion recognition method based on multi-source data fusion is proposed. First, build a multimodal music emotion model, then use TF-IDF to extract lyric modal emotion features, and use Mel frequency cepstrum coefficient to extract audio modal emotion features. Then, after preprocessing the extracted multimodal features, fuse the two multi-source data features of lyric mode and audio mode, and finally calculate the probability distribution of a song in the emotional space according to the fusion results, The emotion category with the highest corresponding value is taken as the emotion category to which the music belongs, so as to achieve the purpose of emotion recognition of multimodal music. Simulation results show that the proposed method has higher accuracy and shorter recognition time for multimodal music emotion recognition.
    Keywords: multi-source data fusion; multimodal music; emotional recognition; inverse text frequency IDF; Mel frequency cepstrum coefficient; MFCC.
    DOI: 10.1504/IJRIS.2023.10056196
     
  • A visual enhancement method for art works based on wavelet packet decomposition   Order a copy of this article
    by Jingjian Zhou 
    Abstract: Aiming at the problems of traditional visual enhancement methods for art works, such as large error of image visual high-frequency feature decomposition, poor brightness enhancement effect and long enhancement time, a visual enhancement method for art works based on wavelet packet decomposition is proposed. First, determine the receptive field of human visual system, then collect the image brightness information through the human visual receptive field, extract the visual information of art works image with Fechner’s logarithm law, finally decompose and reconstruct the high-frequency features of visual information with wavelet packets, and construct an image enhancement model through LIP model to obtain the enhancement results. The experimental results show that the proposed method can reduce the image VHF feature decomposition error, and the enhancement effect is good, and the enhancement time is short.
    Keywords: wavelet packet decomposition; images of art works; visual enhancement; LDA algorithm; normalised complementary transformation.
    DOI: 10.1504/IJRIS.2023.10056197
     
  • Global Path Planning of Mobile Robot Based on Adaptive Sampling Area RRT   Order a copy of this article
    by Zhixiang Hou, Xiao TIAN, Fengling LI, Liefeng HU 
    Abstract: This paper proposed an adaptive sampling area RRT (ASA-RRT) algorithm. Firstly, this paper designs an adaptive sampling region method, which determines the next sampling interval according to the new nodes generated each time, and gradually shrinks the sampling region to reduce sampling in unnecessary regions. Secondly, the target point guidance and greedy strategy are used to strengthen the directionality and convergence speed of the algorithm expansion. Finally, the reverse optimisation algorithm is used for post-processing to eliminate the redundant nodes in the path and shorten the path length. In the simulation environment, the ASA-RRT algorithm is applied to the path planning of the mobile robot. The simulation results show that compared with the basic RRT and target deflection RRT algorithms, the search time, number of sampling points and long path length of the ASA-RRT algorithm are lower by74.05%, 89.08% and 22.45%; 32.67%, 65.07% and 15.61%.
    Keywords: mobile robot; path planning; RRT algorithm; adaptive sampling area.
    DOI: 10.1504/IJRIS.2023.10056198
     
  • Research on Fast Recognition of Athletes' Abnormal Mood Based on Improved MEDA Algorithm   Order a copy of this article
    by Pin Lv, Haixin Huang 
    Abstract: In order to improve the accuracy and average recognition rate of athletes’ abnormal emotions and reduce the recognition time, a fast recognition method of athletes’ abnormal mood based on improved MEDA algorithm was proposed. Firstly, mathematical morphology is used to refine the edge of athlete’s abnormal emotion image, and the edge isolated points in the image are removed. Secondly, the wolf colony algorithm is used to segment different regions of the image, and the rotation correction method is used to extract the region of interest of the image. Finally, the improved MEDA algorithm is used to effectively select the features of the region of interest of the image, and the abnormal emotion type is judged by combining the regional feature screening results. The experimental results show that the proposed method has obvious advantages in recognition accuracy, average recognition rate and recognition time, and the recognition effect are good.
    Keywords: improved MEDA algorithm; emotional recognition; mathematical morphology; wolf colony algorithm; rotation correction.
    DOI: 10.1504/IJRIS.2023.10056199
     
  • Color Correction Method of Fine Arts and Traditional Chinese Painting Works Based on Color Space Decomposition   Order a copy of this article
    by Xiyang Li, Wenzhen Ku 
    Abstract: Aiming at the problems of low colour correction accuracy and large colour space decomposition error in colour correction of fine arts and traditional Chinese painting works, a colour correction method based on colour space decomposition is designed. First, the linear weighting algorithm is used to fuse the colour pixel level of the fine arts and traditional Chinese painting image, and the wavelet transform algorithm is introduced to complete the extraction of colour features, and then the colour features are pre-processed. Finally, determine the state of colour space, through the decomposition of row and column pixels in colour space, calculate the deviation probability with Markov chain, and realise colour correction research through the determination of state function. The results show that the proposed method can improve the accuracy of colour correction of fine arts and traditional Chinese painting images, and the colour space decomposition error is low.
    Keywords: fine arts and traditional Chinese painting works; image colour; pixel level; wavelet transform algorithm; high frequency component; Markov chain.
    DOI: 10.1504/IJRIS.2023.10056886
     
  • Research on automatic recognition of electronic communication information anomalies based on fuzzy association rules   Order a copy of this article
    by Ying He  
    Abstract: In order to improve the accuracy and efficiency of electronic communication information anomaly recognition, this paper proposes the research of electronic communication information anomaly automatic recognition based on fuzzy association rules. Firstly, obtain electronic communication information, and perform attribute extraction, missing value filling, data de-duplication, and data synchronisation. Secondly, the membership matrix of the sample set is constructed, and the fuzzy grade of communication information is determined by fuzzy association rules to achieve the classification of electronic communication information anomaly types; then, K-means++ algorithm is used to optimise the selection of initial centroid; finally, K-FKNN is used to complete the automatic identification of electronic communication information anomalies. The experimental results show that the anomaly information capture rate of this method is 100%, the recognition accuracy is 98%, and the recognition time is only 16 s.
    Keywords: Lagrange multiplier method; fuzzy association rules; membership matrix; fuzzy grade; K-means++ algorithm.
    DOI: 10.1504/IJRIS.2024.10056992
     
  • Student emotion classification based on online comment text big data mining   Order a copy of this article
    by Zhiling Yang 
    Abstract: In order to solve the problems of low recall and accuracy in traditional student emotion classification methods, and long time consuming in student emotion classification, a student emotion classification method based on online comment text big data mining was proposed. Firstly, Selenium is used to crawl the online review text big data, and mean shift clustering algorithm is used to cluster and mine the crawled data. Secondly, logical regression model and sorting neighbour method are used to complete missing data filling and duplicate data cleaning. Finally, the topic model is used to extract the subject words from the online review text data, and calculate the characteristics of the online review text data. Combined with XGBoost, students’ emotions are classified. The experimental results show that the maximum classification recall rate of this method is 97%, the maximum accuracy rate is 98%, and the maximum classification time is 0.84 s.
    Keywords: online comments; text big data mining; Selenium; mean shift clustering; XGBoost.
    DOI: 10.1504/IJRIS.2024.10056994
     
  • Research on user personalized product recommendation method of e-commerce platform based on rough set   Order a copy of this article
    by Xin Yao, Xiaowei Ma, Shizhong Guo 
    Abstract: To improve the purchase rate and recommendation accuracy of personalised product recommendation methods, this paper proposes the research of personalised product recommendation methods for e-commerce platform users based on rough sets. First of all, extract the recommended product features, and use naive Bayesian method to classify the emotion of product features; Secondly, based on rough set theory, the similarity of product recommendation is calculated and the product similarity is arranged; Then, build the scoring matrix to design the product recommendation function; Finally, the product popularity is introduced to score and predict the target users, so as to generate the final Top-N product list and realise personalised product recommendation for users. The results show that the purchase rate of recommended products users in this method is between 68.3% and 88.0%, and the accuracy of product recommendation can reach 0.989, effectively improving the effect of personalised product recommendation.
    Keywords: rough set; similarity; weighted summation; Naive Bayes; commodity popularity.
    DOI: 10.1504/IJRIS.2024.10056995
     
  • Railway wireless train communication network scheduling based on GA-BP algorithm   Order a copy of this article
    by Chenyou Li, Zhinan Li, Yafei Li 
    Abstract: Aiming at the problems of high packet loss rate, large scheduling error and long scheduling delay in the railway wireless train communication network scheduling. Therefore, a GA-BP-based scheduling algorithm for railway wireless train communication network is proposed. Firstly, the structure characteristics of railway wireless train communication network are analysed; then complete the determination of dispatching technical parameters such as line impedance value, reflection coefficient, impedance value during line signal transmission, communication period, etc. Finally, the GA-BP scheduling model is constructed by combining the genetic algorithm and BP neural network algorithm, and the fitness function is introduced to modify the scheduling results to realise the scheduling algorithm design. The test results show that the packet loss rate in the proposed method is low, and the scheduling error is small.
    Keywords: GA-BP; railway wireless train; communication network; scheduling algorithm; technical parameters; fitness function.
    DOI: 10.1504/IJRIS.2024.10056996
     
  • Multi-scale detail enhancement method of 2D animation image based on Retinex algorithm   Order a copy of this article
    by Shu Xin 
    Abstract: In order to solve the problems of low accuracy and long time cost in image multiscale detail enhancement, a method of 2D animation image multiscale detail enhancement based on Retinex algorithm is proposed. Firstly, multi-scale detail pixels are segmented by weighted undirected graph method, and the segmentation results are optimised with the help of edge contour and texture feature information. Secondly, the noise value in detail pixels is obtained and removed by Gaussian decomposition, and the colour vector value in pixel chromaticity space is obtained. Finally, the illumination component is estimated with the help of Gaussian convolution function, and the enhancement model based on Retinex algorithm is constructed to achieve detail enhancement. The test results show that the proposed method can improve the detail enhancement accuracy, with the enhancement accuracy up to 98%, and the enhancement time is shortened.
    Keywords: Retinex algorithm; 2D animation image: multi-scale details; enhancement method; transformation function slope.
    DOI: 10.1504/IJRIS.2024.10056998
     
  • Dynamic obstacle avoidance method for autonomous mobile robot based on machine vision   Order a copy of this article
    by Hui Li, Shuo Liang, Xiangyu Han 
    Abstract: In order to improve the accuracy of dynamic obstacle avoidance and shorten the path of robot dynamic obstacle avoidance, this paper proposes a new dynamic obstacle avoidance method for autonomous mobile robots based on machine vision. Firstly, a kinematics model of an autonomous mobile robot is constructed to obtain information such as the robot’s motion direction and speed. Secondly, machine vision technology is used to obtain the robot motion trajectory in real-time based on on-site environmental information, and YOLO algorithm is used to segment the robot image obtained by the machine vision camera to detect dynamic obstacles. Finally, DWA algorithm is used to detect whether there are moving obstacles on the robot’s motion trajectory in real-time to achieve dynamic obstacle avoidance for the robot. The experimental results show that the proposed method has a higher obstacle avoidance accuracy, reaching a maximum of 96.7%, and the obstacle avoidance path is significantly shortened.
    Keywords: machine vision; robot obstacle avoidance; YOLO algorithm; DWA algorithm; kinematic model.
    DOI: 10.1504/IJRIS.2024.10056999
     
  • Research on Efficient detection method of low voltage power line based on edge intelligence   Order a copy of this article
    by Ziwen Cai, Yun Zhao, Yong Xiao, Yuxin Lu, Haolin Wang 
    Abstract: In view of the complex background of overhead lines in low-voltage distribution networks in aerial photos, power lines are blurred, and the characteristics of the lines will be seriously weakened. A context based Gabor YOLO algorithm is proposed for efficient detection of low-voltage power lines. Firstly, an improved Gabor operator is proposed to extract Gabor features from greyscale and Gaussian filtered images to obtain the foreground region of the image; Secondly, an improved YOLO neural network model is used to locate and detect power lines and auxiliary targets in the foreground image; and then conduct experimental verification. Experimental results show that this method has the highest accuracy and extraction speed compared to other methods such as YOLOv4, with a mAP value of 93.6%, which meets the actual work needs.
    Keywords: overhead lines; low-voltage; distribution network; Gabor YOLO; edge intelligence.
    DOI: 10.1504/IJRIS.2024.10057000
     
  • Real time Sharing Method of Panoramic Data of Substation Equipment Based on RAFT Algorithm   Order a copy of this article
    by Chengjie Cao, Jiaqi Zheng, Yifei Fan, Fangyuan Tian, Darui He 
    Abstract: In order to solve the problems of low consistency and security in traditional data sharing methods, this paper designs a real-time sharing method of panoramic data of substation equipment based on RAFT algorithm. First, analyse the data generation process, and collect the panoramic data of the device in real time and quickly according to the location and environment of different devices. Then, the physical and logical relationship model between panoramic data is constructed, and the similarity data pair matrix is used to ensure the uniqueness of the data and complete the data preprocessing. Finally, RAFT algorithm is used to copy the data state, determine the data sharing node, and determine the degree of association between panoramic data through the weighted map, so as to achieve real time sharing of panoramic data in the same area. The experimental results show that the security factor and consistency factor can be guaranteed to be about 0.98 in the sharing process.
    Keywords: RAFT algorithm; substation equipment; panoramic data; data sharing; device image.
    DOI: 10.1504/IJRIS.2024.10057001
     
  • Data mining method for English distance learning based on weighted fast clustering   Order a copy of this article
    by Xiaohong Yu  
    Abstract: In order to improve the efficiency of English distance teaching data mining, mining recall and accuracy, a weighted fast clustering based English distance teaching data mining method is proposed. Firstly, the data of English distance teaching are collected from three dimensions: students’ basic characteristics, online teaching behaviour characteristics, and students’ learning effect characteristics; Secondly, data preprocessing is realised through data integration, data selection, data cleaning, attribute construction and other processes; Finally, based on the weighted depth forest, calculate the attribute weight of the data, calculate the similarity between the data through the weighted fast clustering method, determine the English distance teaching data mining list through the similarity between teaching data, and realise the English distance teaching data mining. The experimental results show that the mining accuracy and recall of this method are high, and the mining time is short, indicating that the data mining effect of this method is good.
    Keywords: data cleaning; data preprocessing; weighted depth forest; data mining; weighted fast clustering.
    DOI: 10.1504/IJRIS.2024.10057002
     
  • Intrusion risk detection method of power network based on dynamic correlation analysis   Order a copy of this article
    by Yunhao Yu, Fuhua Luo, Xiang Guo 
    Abstract: Aiming at the low accuracy, recall rate, and F1 value of traditional intrusion risk detection methods, a dynamic association analysis based intrusion risk detection method for power networks is proposed. Firstly, the network intrusion detection data is normalised using the max-min method. Based on the data normalisation results, the power network intrusion feature dimensionality is reduced using the PCA-ReliefF method. Secondly, based on the dimensionality reduction results of network intrusion features, a dynamic association analysis method is used to calculate the specific weights of risk nodes, and the calculation results are graded to obtain network intrusion risk assessment results. Finally, based on the network intrusion risk assessment results, an artificial immune method is used to detect the power network intrusion risk. Experimental results show that the intrusion risk detection accuracy, recall rate, and F1 value of this method have been significantly improved.
    Keywords: dynamic correlation analysis; power network; intrusion risk; PCA-ReliefF; artificial immune method.
    DOI: 10.1504/IJRIS.2024.10057003
     
  • Collaborative filtering recommendation algorithm for e-commerce products based on bayesian network   Order a copy of this article
    by Hongli Wan 
    Abstract: In order to effectively improve the recommendation accuracy of collaborative screening of e-commerce products, ensure the results of collaborative screening of e-commerce products, and reduce the time consumption of collaborative screening of e-commerce products, this project plans to study the recommendation algorithm for collaborative screening of e-commerce products based on Bayesian networks. Use an identical method to determine the user selection of e-commerce platform. On this basis, a new method was adopted to conduct an overall similarity analysis of products in e-commerce platforms. Then, using Bayesian networks and expected maximum methods, establish a model of user interest in e-commerce products on e-commerce platforms, thereby completing joint screening of e-commerce products. Through testing the joint screening of different types of e-commerce products, it was found that this method had a recommendation rate of 91.6% and an accuracy rate of 95.9% in the joint screening of e-commerce products.
    Keywords: Bayesian network; E-commerce products; expectation maximisation algorithm; collaborative filtering recommendation; slope one algorithm.
    DOI: 10.1504/IJRIS.2024.10057004
     
  • A method of removing redundant information from multidimensional data based on bayesian algorithm   Order a copy of this article
    by Chunlei Ren, Yanhua Wu, Demin Ma 
    Abstract: In order to solve the problems of poor anti-interference, high false detection rate of redundant information features, and high time cost of redundant removal in traditional methods, this paper proposes a method of removing redundant information from multidimensional data based on Bayesian algorithm. Firstly, noise information and interference information in multidimensional data are filtered; secondly, the preprocessed data is segmented to extract the features of redundant information; then, the fitness function of redundant information is used to classify redundant information features; finally, the association model between nodes is established in the Bayesian network, and the redundant information is filtered using the Hash function. The experimental results show that the design method has good anti-jamming performance, the global maximum false detection rate is 5.29%, and the maximum value of the time cost to remove redundancy is 213.06 s, which shows that the method has good application performance.
    Keywords: multidimensional data; redundant information; feature extraction; feature classification; Bayesian algorithm; associative model.
    DOI: 10.1504/IJRIS.2024.10057005
     
  • Stability detection method of large-scale network information transmission based on digital twin technology   Order a copy of this article
    by Xiao Lin, Jian Du, Wenjun Gao, Aobo Zhou 
    Abstract: In order to overcome the problems of low accuracy of transmission channel state estimation, low detection accuracy and long detection task completion time in traditional transmission stability detection methods, a large-scale network information transmission stability detection method based on digital twin technology is proposed. The data twin technology is used to build the network data acquisition architecture, extract the analysis results of large-scale network information transmission channel characteristics, and obtain the transmission channel state estimation results with the channel state correlation coefficient matrix. The output of the detection model is adjusted according to the fuzzy logic to obtain accurate detection results. The experimental results show that the average correct rate of channel state estimation of large-scale network information transmission is 96.9%, the detection accuracy of large stability is high, and the maximum detection task completion time is 2.0 s. The practical application effect is good.
    Keywords: digital twin technology; large-scale; network information transmission; stability test; channel state estimation; fuzzy logic.
    DOI: 10.1504/IJRIS.2024.10057006
     
  • A personalized recommendation method of online and offline mixed teaching resources based on user preference behavior   Order a copy of this article
    by Qingqing He 
    Abstract: For online and offline hybrid teaching resources, due to the low recommendation accuracy and long recommendation time of traditional personalised recommendation methods, a personalised recommendation method based on user preference behaviour is proposed. First, we collect the teaching resource data through the crawler technology, and then clean the obtained data, and then build the teaching resource model. Finally, we build the user model, calculate the interest preference behaviour group category that the user belongs to, determine the user preference behaviour, and use cosine similarity to measure the similarity between users, so as to predict the user score, and recommend the resource with the highest score to the user. The experimental results show that the proposed method has higher accuracy and shorter recommendation time.
    Keywords: user preference behaviour; teaching; personalised recommendation; crawler technology; cosine similarity.
    DOI: 10.1504/IJRIS.2024.10057007
     
  • A coverage enhancement method for wireless sensor networks based on firefly algorithm   Order a copy of this article
    by Qiuju Zhang 
    Abstract: In order to improve the coverage enhancement effect and balance of wireless sensor networks, and shorten the enhancement time, this paper proposes a coverage enhancement method of wireless sensor networks based on firefly algorithm. By describing the coverage problem of wireless sensor network, firefly algorithm is used to deploy wireless sensor network nodes. Through the reasonable deployment of nodes and the shutdown of redundant nodes, the coverage enhancement model of wireless sensor networks is established to achieve the best resource allocation. The experimental results show that the coverage enhancement effect and balance of the proposed method are good, and the coverage enhancement time of wireless sensor networks can be shortened. The highest coverage rate of this method is 99.1%.
    Keywords: firefly algorithm; wireless sensor network; network coverage enhancement; node deployment; network coverage.
    DOI: 10.1504/IJRIS.2024.10057008
     
  • Forecasting DDoS Attack with Machine Learning for Network Forensic Investigation   Order a copy of this article
    by Saswati Chatterjee, Suneeta Satpathy, Bijay Paikaray 
    Abstract: The recognition of intrusion attempts is the fundamental region of network security, with the objective of identifying the impact of these actions on the distinctive variations of the captured traffic. Innovation and inquiry are now necessary for most of the attacks. In this paper, a machine-learning approach has been used to track anomalous network traffic. Also, the statistical parameters are employed to enhance the performance based on learning models. The paper also represents a detection pattern through machine learning experiments for DDoS attacks on KDD Cup 99 dataset where ten features are used for accuracy measures. The study further classifies the hidden forms to sense the DDoS attack patterns. The paper further concludes with the experimental outcomes that establish the improved performance assessment of the K-nearest neighbour algorithm in comparison to other predictable learning approaches and thus the proposed model can reach the uppermost correctness.
    Keywords: distributed denial of service; DDoS; K-nearest neighbour; KNN; classification; machine learning; support vector machine; SVM; true positive rate; TPR.
    DOI: 10.1504/IJRIS.2024.10057010
     
  • Study on classification and retrieval method of e-book resources based on binary sort tree   Order a copy of this article
    by Liping Zhong 
    Abstract: In order to solve the problems of low classification accuracy and large retrieval error in the classification and retrieval of e-book resources, a classification and retrieval method of e-book resources based on binary sorting tree is designed. Calculate the similar distance between e-book resource data and realise the structural analysis of e-book resource platform. The vector space model is introduced to realise the data integration of e-book resources. The retrieval model is constructed and the stability coefficient is introduced to realise the research. The experimental results show that the classification accuracy of the proposed method is about 99%, the retrieval error is always less than 1%, and the time cost is less than 2 seconds, which has certain advantages.
    Keywords: binary sort tree; BST; E-book resources; search model; Vector space model; similar distance.
    DOI: 10.1504/IJRIS.2023.10057741
     
  • Separating voice and background music based on 2DFT transform   Order a copy of this article
    by Maoyuan Yin, Li Pan 
    Abstract: A new separation method for human voice and background music is designed to address the problems of large positioning errors, large feature extraction errors, and low separation accuracy in existing methods. Firstly, a microphone array is setup in the virtual space to complete signal denoising, and a generalised cross correlation function is introduced to achieve signal localisation. Then, construct a signal time spectrum graph, calculate the position change of signal energy on the frequency axis, and extract components in the sound signal frequency band and time frame. Finally, hamming window function is introduced to improve the 2DFT transform algorithm and build a signal separation model. The test results show that when the proposed method is applied, the localisation error of human voice is only 0.50% when the frame rate of human voice in audio is 1,000 kbps, and the error of background music feature extraction is only 0.05% when the sample audio sampling rate is 60 KHz. The separation accuracy of human voice and background music remains above 95%, with a maximum of nearly 99%. The application effect is good.
    Keywords: 2DFT; voice; background music; separation; generalised cross correlation function; microphone array; separation model.
    DOI: 10.1504/IJRIS.2023.10057742
     
  • Hospital infant emergency and critical information integration based on improved fuzzy clustering   Order a copy of this article
    by Juan Xiao, Xiaoli Liu, Jina Zhang 
    Abstract: In order to solve the problems of low recall rate, high integration error and long time consuming in traditional information integration methods, a hospital infant emergency and critical information integration method based on improved fuzzy clustering is designed. Firstly, Manhattan distance is introduced to measure the degree of correlation between information, and the information feature is extracted by mutual information method. Then, the information feature missing pattern is determined, and the KNN algorithm and MisForest interpolation algorithm are used to fill the feature missing. Finally, based on the filling of feature missing, the fuzzy clustering algorithm is improved according to the membership degree of feature information, and the improved algorithm is used to realise the integration of hospital infant’s emergency and critical information. The experimental results show that the proposed method has high recall rate, low integration error and short time consumption.
    Keywords: improve fuzzy clustering; hospital infant; emergency and critical; information integration; KNN algorithm; MisForest interpolation algorithm.
    DOI: 10.1504/IJRIS.2023.10057743
     
  • Intelligent retrieval method of book resources in smart library based on RFID   Order a copy of this article
    by Wei Huang, Jing Ling 
    Abstract: In view of the problems existing in the existing methods such as large retrieval error, low data positioning accuracy, and long retrieval time, this paper proposes an intelligent retrieval method for book resources in smart libraries based on RFID. First of all, complete the rough extraction of the features of the book resources in the smart library. Then, set the length rule of book resource features, eliminate ambiguous data, determine the similarity of book resource features, and merge similar features through similar probability mapping. Finally, complete the feature data authentication, set the tag with RFID tag technology, locate the tag data location, build the book resource index tree, and determine the intelligent retrieval model. The test results show that the proposed method can reduce the intelligent retrieval error of book resources in smart libraries, improve the accuracy of data location, and reduce the retrieval time cost.
    Keywords: RFID; smart library; book resources; intelligent retrieval; label end; participle dictionary.
    DOI: 10.1504/IJRIS.2023.10057744
     
  • Fuzzy vision image denoising algorithm based on Bayesian estimation   Order a copy of this article
    by Cong Ye 
    Abstract: A fuzzy visual image denoising algorithm based on Bayesian estimation is proposed to address the problems of poor denoising performance and long denoising time in traditional image denoising algorithms. Analyse the noise reduction requirements of fuzzy visual images, construct a visual image degradation model, and obtain fuzzy visual images; Analyse the spectrum of the acquired fuzzy visual image, preprocess the spectrum signal by adding a window function, and detect the edge of the fuzzy visual image according to the preprocess result to supplement the fuzzy edge; Based on the supplemented blurred visual image edges, Bayesian estimation is used for image denoising to improve the image denoising effect. The experimental results show that the visual image obtained by the proposed method is relatively clear and has high image quality, indicating that the denoising effect of the algorithm is good and the denoising time is short.
    Keywords: Bayesian estimation; blur visual image; spectrum analysis; window function; blur edges.
    DOI: 10.1504/IJRIS.2023.10057745
     
  • Reflective Clothing Detection Based on YOLOv5s Fused With SE Attention Mechanism   Order a copy of this article
    by Yuanyuan Wang, Dingyun Gu, Yemeng Zhu, Zhihan Zhang, Jiahui Cao, Haiyan Zhang 
    Abstract: This paper proposes a YOLOv5s deep learning algorithm incorporating the SE attention mechanism to address the issue of workers failing to wear reflective clothing on duty, which has resulted in casualties from time to time. The YOLOv5s model is used to train the reflective clothing dataset obtained by collecting images of construction sites, and the SE attention mechanism module is added to the network structure of YOLOv5 for improved performance. The reflective clothing detection algorithm is then deployed to the web for real-time detection. The experimental results show that the algorithm achieves an mAP value of 96.70% with the SE attention module, resulting in an accuracy improvement of 1.09% compared to the model without the SE module. This enables practical applications to reduce the incidence of construction accidents at construction sites.
    Keywords: reflective clothing detection; YOLOv5; attention mechanism; vue; flask.
    DOI: 10.1504/IJRIS.2023.10057746
     
  • Fast classification algorithm of uncertain big data stream based on distributed limit learning machine   Order a copy of this article
    by Zhiling Yang 
    Abstract: Aiming at the problems of low classification accuracy and long classification time in traditional uncertain big data stream classification algorithm, a fast classification algorithm for uncertain big data stream based on distributed limit learning machine is proposed. First, construct the uncertain big data model, and measure the adaptive mixed distance of the data flow. Then, according to the measurement results, use the DSUFIM ? mine mining algorithm to mine the uncertain big data flow. Then combine the K-nearest neighbour and label correlation to conduct online feature screening for the mined uncertain big data flow, and use the cosine similarity method to detect whether the feature has conceptual drift. Finally, according to the detection results, the distributed limit learning machine is used to quickly classify uncertain big data streams. The experimental results show that the proposed algorithm has higher classification accuracy and faster classification speed for uncertain large data streams.
    Keywords: distributed limit learning machine; uncertain big data flow; quick classification; K nearest neighbour; cosine similarity.
    DOI: 10.1504/IJRIS.2023.10057747
     
  • Balanced Scheduling Method of Ideological and Political Teaching Resources Based on Cluster Analysis Algorithm   Order a copy of this article
    by Wenbin Liu 
    Abstract: In order to overcome the problems of poor scheduling accuracy, low recall rate and long scheduling time in the traditional scheduling method of ideological and political resources, the paper proposes a balanced scheduling method of resources based on clustering analysis algorithm. First, the load time series of the teaching resource server is determined and pre-processed. Secondly, cluster analysis is used to classify the data. Finally, according to the classification results, the balanced scheduling function of resources is constructed, and the particle swarm optimisation algorithm is used to solve the scheduling function to obtain the final scheduling strategy. The results show that the scheduling accuracy of the proposed method is 99.12%, the recall rate is up to 95%, the scheduling time is controlled within 7 s, and the resource balance scheduling effect is good.
    Keywords: clustering analysis algorithm; balanced scheduling; time series; particle swarm optimisation.
    DOI: 10.1504/IJRIS.2023.10057748
     
  • UAV patrol detection of forest fire burning point based on YOLOv3-SPP algorithm   Order a copy of this article
    by Lishan Ma, Yuanjie Ding, Shunhu Dong, Wenjun Chen, Shenghong Wang 
    Abstract: To overcome the problems of low recall rate, poor positioning accuracy and long time of UAV patrol inspection of forest fire location detection method, this paper proposes a UAV patrol inspection method of forest fire burning point location based on YOLOv3-SPP algorithm. Firstly, the forest fire images are collected by UAV patrol technology. Then, the two-dimensional Gaussian function is used for image filtering to realise fire image processing. Finally, YOLOv3 target detection network is constructed, image feature fusion is carried out with SSP method, and the sigmoid detection function of burning point position is constructed and solved to realise the detection of burning point position of forest fire. The results show that the detection recall rate of this method is 98%, the positioning accuracy is 98%, and the detection time is only 9.2 s, indicating that this method can effectively improve the detection effect of fire burning point location.
    Keywords: YOLOv3; two-dimensional Gaussian function; SPP; forest fire.
    DOI: 10.1504/IJRIS.2023.10057749
     
  • A personalized recommendation of tourism routes based on rule-based reasoning   Order a copy of this article
    by Qiang Wu, JiaHui Peng, DongMei Ge 
    Abstract: In order to solve the problems of low accuracy, long time and low satisfaction of tourists existing in traditional methods, this paper proposes a personalised recommendation method of tourism routes based on rule-based reasoning. Firstly, the web crawlers is used to collect travel data and process the collected data dimensionless. Then, the rule reasoning method is used to build the personalised preference model of tourists, and the random gradient rise method is used to optimise the model. Finally, determine the constraints, take the personalised information of tourist routes, scenic spots and tourists as the input vector of the model, build the personalised recommendation model of tourist routes, and get the relevant recommendation results. The experimental results show that the maximum recommendation accuracy of this method is 96%, the recommendation time is always less than 51 ms, and the average tourist satisfaction is 9.70.
    Keywords: rule reasoning; tourist routes; personalised recommendation; web crawler; preference model; constraint condition.
    DOI: 10.1504/IJRIS.2023.10057750
     
  • Study on Node preserving routing coverage algorithm for sensor networks based on compressed HMAC.   Order a copy of this article
    by Jiwei Li 
    Abstract: Research on node maintenance routing coverage algorithms in sensor networks can improve node coverage and node activity in sensor networks. This paper proposes a node preserving routing coverage algorithm for sensor networks based on compressed HMAC. Firstly, based on the VGG-F network structure, high-dimensional features of the sensor network layer are extracted to maximise the amount of information carried by the hash code; Secondly, a feature extractor is designed using compressed HMAC algorithm; Then, uniform coverage of the target area is achieved according to the force balance conditions of the acting nodes; Finally, the compressed HMAC method is used to achieve target node division, and the coverage algorithm is used to optimise node retention routing and mobile coverage. The experimental results show that the coverage rate of this method can reach up to 99%, and the node utilisation rate is high.
    Keywords: feature extractor; compressed HMAC; Hash code; VGG-F network structure; regularisation term.
    DOI: 10.1504/IJRIS.2023.10057751
     
  • FUZZY EXPERT SYSTEM FOR ACCESS CONTROL OF CHILDREN TO THE INTERNET   Order a copy of this article
    by Rasim Alguliyev, Fargana Abdullayeva, Sabira Ojagverdiyeva 
    Abstract: As children’s use of the internet increases, serious online safety issues arise. As a result, it is observed that the damage to their psychology and health is increasing. Along with harmful content from web pages, the negative impact of constant use of digital devices leaves deep marks on children’s health and psychology. In order to overcome these problems and prevent harm, there is a need to implement programs that control access to the internet, filter harmful content on web pages, constantly monitor children’s behaviour, make assessments and make accurate decisions. The article proposes a method of internet access control using a fuzzy logic inference system. This method is focused on the individual user and is done by imposing restrictions on their use of technologies (computers, phones, tablets, etc.). Screen time is determined for the use of technology, taking into account the age of the user, health and psychological diseases.
    Keywords: fuzzy logic extraction system; FLES; online child safety; malicious information; Mamdani model; screen time; age; children’s health; psychology of children.
    DOI: 10.1504/IJRIS.2023.10057752
     
  • Visual interactive system of cultural communication based on Unity3D   Order a copy of this article
    by Qiao Wu 
    Abstract: In order to improve the system response efficiency of cultural communication visual interaction system, the design and research of cultural communication visual interaction system based on Unity3D technology was carried out. Firstly, the framework design of visual interaction system for cultural communication is determined; Then, according to the relevant parameters, select the mobile mode suitable for different immersive interaction scenarios, and complete the mobile interaction module design; Then, the virtual handwriting module is designed based on the graphics rendering system and OCR technology; Finally, with the help of Unity3D technology, the layout design of the interface of the cultural communication visual interaction system is completed, and the system construction is completed. The experimental results show that the memory occupancy of the system is only 13.5%, and the interactive response time is only 23s. The system application performance is better.
    Keywords: Unity3D; cultural communication; visual interaction; OCR technology; community division; force-oriented model.
    DOI: 10.1504/IJRIS.2023.10057753
     
  • A personalized matching method of English learning data packages based on knowledge map   Order a copy of this article
    by Wei Liu  
    Abstract: In order to improve the retention rate of learning data and the coverage rate of matching results, this paper designs a personalised matching method of English learning data packages based on knowledge map. First, cluster the learning packages information and analyse user interest. Secondly, the knowledge map network is established and the sequence nodes are analysed. Then, the user features and data packages features are decomposed from the relational entities of the knowledge map to obtain independent user features and data packages features; Finally, users’ interests and preferences are optimised by updating the node weights in the knowledge map, and personalised matching is realised according to the similarity between data packets. Experiment shows that this method can control the retention rate of the learning date packages at more than 90% and the matching coverage at more than 95%, which shows that this method can meet the needs of users.
    Keywords: English learning data packages; users; personalised matching; knowledge map; node weight.
    DOI: 10.1504/IJRIS.2023.10057754
     
  • The style feature extraction of electronic music based on kernel limit learning machine   Order a copy of this article
    by Minyuan Jiang  
    Abstract: In order to solve the problems of low recall, low signal-to-noise ratio and low extraction accuracy of traditional methods, a style feature extraction method of electronic music based on kernel limit learning machine was proposed. Firstly, build an electronic music signal acquisition architecture, and process the acquired signal by framing, median smoothing, and three-point linear smoothing. Then, wavelet transform and LMS algorithm are used to build an adaptive filtering model for electronic music signal, and the smoothed signal is denoised. Finally, the model of electronic music style feature extraction based on kernel limit learning machine is built, and the signal denoising results are input into the trained model to get the results of electronic music style feature extraction. The experimental results show that the maximum recall is 98.3%, the maximum SNR is 57.6dB, and the extraction accuracy is always above 95.1%.
    Keywords: Kernel limit learning machine; electronic music; style characteristics; feature extraction; framing; median smoothing; three point linear smoothing.
    DOI: 10.1504/IJRIS.2023.10057755
     
  • Automatic Fusion Method for Perceptual Data of Internet of Things Based on Kalman Filter   Order a copy of this article
    by Andi Gao, Xiaojing Guo, Ketong Liu, Kuntai Meng 
    Abstract: Due to the low fusion accuracy of traditional internet of things (IoT) sensing data automatic fusion methods, an automatic fusion method of IoT sensing data based on Kalman filter is proposed. First of all, we adopt the error correction mechanism and time registration to deal with the structure deviation of the IoT sensing data and the asynchronous problem of the IoT sensing data; then, the Kalman filtering algorithm is used to fuse the time data and space data at the terminal node and gateway layer. Finally, the Kalman filtering algorithm is optimised by using the distribution diagram method to determine the abnormal or missing data in the fused data, so as to obtain the optimised data automatic fusion results. The test results show that the precision of this method for automatic fusion of IoT sensing data is above 94%, and the fusion effect is good.
    Keywords: Kalman filter; internet of things; IoT; perception data; automatic fusion.
    DOI: 10.1504/IJRIS.2023.10057756
     
  • A Personalized recommendation algorithm of ideological and political education resources based on hybrid collaborative filtering   Order a copy of this article
    by Shan Zhang 
    Abstract: In order to overcome the problems of large error and long time consuming of existing recommendation methods, this paper proposes a personalised recommendation algorithm of ideological and political education resources based on hybrid collaborative filtering. First, with the help of Oracle, collect ideological and political education resources data to complete the data extraction of ideological and political education resources. Secondly, the data of ideological and political education resources are classified, data removal and data dimension reduction pre-treatment. Finally, the recommended user-resource evaluation matrix of ideological and political education resources is constructed to determine the similarity of the recommended objects of ideological and political education resources. Hybrid collaborative filtering algorithm is introduced to build a personalised recommendation model and realise the personalised recommendation. The experimental results show that the method can reduce the recommendation error and shorten the recommendation time, with the highest recommendation error exceeding 0.03%.
    Keywords: hybrid collaborative filtering; ideological and political education resource data; Oracle; personalised recommendation algorithm.
    DOI: 10.1504/IJRIS.2023.10057768
     
  • Study of the Crossing Number Associated with Strong Product of Path with Cycle and Triangular Snake Graph   Order a copy of this article
    by Mhaid Mhdi Alhajjar, Amaresh Chandra Panda, Siva Prasad Behera 
    Abstract: In 2018, Ouyang et al. presented the first efforts related to the crossing number of strong product of the path $P_{m}$ to the cycle $C_{n}$. They proved that $cr(P_{2}$$boxtimes$$C_{n})=n$ for $ngeq3$ together with introducing a general conjecture as follows: $cr(P_{m}$$boxtimes$$C_{n})=(m-1)n$ : $ forall$ $m,ngeq3$ . Here, we prove that Ouyang et al. conjecture is also true for $n=3$ and $mgeq3$ , together with exhibiting an optimal drawing of it. Furthermore, we start to study new case in relation to the strong product of path with triangular snake graph $TS_{n}$ by proving that $cr(P_{2}boxtimes TS_{n})=3lfloordfrac{n}{2}rfloor$ for $ngeq3$ .
    Keywords: Crossing Number; Strong Product; Counting Argument.
    DOI: 10.1504/IJRIS.2023.10057991
     
  • Study on Intelligent travel route recommendation method based on popularity of interest points   Order a copy of this article
    by DongMei Ge, Qiang Wu, ZhiZhu Lai 
    Abstract: In order to improve the accuracy of tourist route recommendation and tourist satisfaction, an intelligent route recommendation method based on the popularity of interest points is proposed. From the perspective of popularity of tourist attractions, tourist travel time and scenic spots travel time, the tourist attractions are scored, and the tourist attractions that meet the needs of tourists are mined according to the score results. Design the POI correlation diagram of tourists’ interest points to obtain tourists’ interest preferences from the perspective of time and preference degree. Considering the popularity of POI, UPST-TB algorithm is used to integrate the interest tag data, and combined with the content-based recommendation idea to realise the intelligent recommendation of tourism routes. The experimental results show that the proposed method effectively improves the accuracy and recall rate of tourism route recommendation results, and can recommend scientific and reasonable tourism routes for tourists.
    Keywords: popularity of interest points; tourist routes; intelligent recommendation; directed weighted graph; UPST-TB algorithm.
    DOI: 10.1504/IJRIS.2023.10057992
     
  • Evaluation of the Shortest Path by Using Bellman Ford's Algorithm in a Fermatean Neutrosophic Environment   Order a copy of this article
    by Prasanta Kumar Raut, Siva Prasad Behera 
    Abstract: The shortest path problem is an extremely important topic in graph theory with numerous real-life applications in science and technology. In this paper, we discuss an improved edition of Bellman-Ford’s algorithm for evaluating the shortest path on a connected network with respect to fermatean neutrosophic numbers as the vertex weights of the given network, which is the elongation of a neutrosophic number. It is straightforward to evaluate the shortest path when the environment is certain, but in an uncertain environment, it is difficult, so here we used the fermatean neutrosophic number. Finally, we implemented our suggested methodology with a mathematical example and finally studied a comparative analysis with respect to the dynamic programming approach and with different types of existing algorithms, and discussed the advantages of our methods.
    Keywords: Bellman’s algorithm; directed graph network; fermatean neutrosophic numbers; FNNs; score functions; shortest path problem; SPP.
    DOI: 10.1504/IJRIS.2023.10057993
     
  • A Method for Predicting Learning Achievements in Online Education Based on Behavioral Characteristics   Order a copy of this article
    by Yan Zhang 
    Abstract: Aiming at the problems of low prediction accuracy, large error in extracting behavioral feature data, and long prediction time in traditional online education learning achievement prediction methods, a behavioral feature based online education learning achievement prediction method is proposed. Firstly, the three-dimensional S-F-T model is used to divide and analyze the students' online education and learning behavior. Then, according to the analysis results, the random forest algorithm is used to calculate the weight of behavior characteristics data, and select the characteristics of students' online education and learning behavior; Finally, a prediction function is constructed through the local self attention mechanism, and the selected network education learning behavior feature data is input as samples into the constructed prediction function, outputting the final prediction result. The test results show that the proposed method has good prediction accuracy, low error in extracting behavioral features, and short prediction time.
    Keywords: Behavioral characteristics; Online education learning; Performance prediction; S-F-T model; Random forest; Local self attention mechanism.
    DOI: 10.1504/IJRIS.2023.10058349
     
  • Personalized recommendation method of educational resources for ideological and political courses based on data mining   Order a copy of this article
    by Jun-cheng Wang, Lin Zeng 
    Abstract: In order to improve the teaching quality of ideological and political courses, a personalized recommendation method of educational resources for ideological and political courses based on data mining is proposed in this paper. Firstly, the topological structure for course resources distribution is established, and the data structure features of educational resources are rearranged through sparse and discrete dimension detection. Then, a personalized recommendation model of educational resources is established to control and optimize the recommendation process of teaching resources. The knowledge map model is used to realize the information interaction between users and projects, and the entity embedding and high-order preference dissemination recommendation methods are used to realize the personalized recommendation of educational resources for ideological and political courses. The experimental results show that this method can contributes to high recommendation satisfaction and has a certain application value.
    Keywords: Data mining; Ideological and political courses; Educational resources; Personalized recommendation; Knowledge transfer.
    DOI: 10.1504/IJRIS.2023.10058350
     
  • Color enhancement processing of interior scene design images based on guided filtering   Order a copy of this article
    by Shashan Hou 
    Abstract: Aiming at the problems of traditional image color enhancement methods, such as low average gradient, low standard deviation after enhancement and low information entropy, an image color enhancement processing method based on guided filtering was proposed for interior scene design. Firstly, a guided filtering algorithm is used to remove the noise in the interior scene design image, and a gradient domain guided filtering algorithm is used to extract the illumination component. Secondly, the correction weights are linearly stretched to different degrees to obtain adaptive gamma correction weights of different gray levels. Finally, color compensation and enhancement of interior scene design images are achieved through gamma correction. The experimental results show that when the number of images is 6000, the information entropy of the color enhanced image in this method is 921 bits, the image standard deviation reaches 4.83, and the average gradient of the image is 23.8.
    Keywords: Guided filtering; Interior Scene Design Image; Image color enhancement; Gamma correction; Color compensation.
    DOI: 10.1504/IJRIS.2023.10058351
     
  • A method of swimmer emotion recognition based on sequence annotation model   Order a copy of this article
    by Yangjun Liu, Junying Yang 
    Abstract: In order to improve the accuracy of swimmer emotion recognition and shorten the time spent on emotion recognition, this paper proposes a swimmer emotion recognition method based on sequence annotation model. Firstly, the original face image of a swimmer is collected, and the collected face image is integrally processed. The original face image is restored through linear equations to obtain a clear face image of the swimmer; Secondly, a face image conversion marker matrix is constructed to match the key points and singular values of the image to extract the features of the swimmer's face image; Then, a sequence annotation model is used to classify the expression features of swimmers; Finally, using fuzzy kernel discriminant analysis technology to complete the swimmer's emotional recognition. Experimental results show that this method can accurately recognize swimmers' facial expressions, with a recognition accuracy of up to 100%.
    Keywords: sequence annotation model; Emotional recognition; Face image; Singular value matching; Swimmers.
    DOI: 10.1504/IJRIS.2023.10058352
     
  • Retrieval Method of Network Education Resources Based on Associated Data   Order a copy of this article
    by Guojuan Li 
    Abstract: Aiming at the problems of low retrieval accuracy and long retrieval time in traditional online education resource retrieval methods, a method of online education resource retrieval based on associated data is proposed. First, establish a model of online education resources, obtain online education resource data through web crawlers, and then preprocess the obtained online education resource data through a series of steps. Then, based on the preprocessing results, construct a semantic grid computing model to extract semantic features of online education resources. Finally, based on semantic features, associate the semantic data of online education resources through semantic similarity and semantic correlation, Based on the associated data, a retrieval model for online education resources is established using naive Bayesian methods to obtain the retrieval results. Simulation results show that the proposed method has higher retrieval accuracy and shorter retrieval time for online education resources.
    Keywords: Associated data; Network education; Resource retrieval; Web crawler; Semantic grid; Naive Bayes.
    DOI: 10.1504/IJRIS.2023.10058353
     
  • A Method of Integrating English Listening and Speaking Distance Teaching Resources Based on Data Deduplication   Order a copy of this article
    by Wei Liu 
    Abstract: The integration of teaching resources can improve students' English listening and speaking abilities. Therefore, a method of distance teaching resources integration based on data deduplication is proposed. Firstly, construct a teaching feature extraction table and collect English listening and speaking distance teaching resources. Secondly, CAONT is introduced to conduct a round of transformation of distance learning resources for English listening and speaking abilities to achieve safe and deduplication processing of teaching resource data. Finally, after completing the deduplication process, a resource integration function for distance learning of English listening and speaking ability is constructed. Through the constraints of temporal relationships between tasks, the resource integration function for distance learning of English listening and speaking ability is solved to achieve the integration of English teaching resources. The experimental results show that the accuracy rate of integrating English teaching resources can reach 98%.
    Keywords: Data deduplication; CAONT; Temporal relationship constraints; Integration of teaching resources;content of courses.
    DOI: 10.1504/IJRIS.2023.10058354
     
  • Digital Ideological and Political Education Data Sharing Algorithm Based on Federal Incremental Learning   Order a copy of this article
    by Aihua Mo 
    Abstract: Traditional sharing methods have problems such as low safety factor, low recall rate, and low efficiency in data sharing. A federated incremental learning based data sharing algorithm for digital ideological and political education is proposed. Based on the data extraction standards of digital ideological and political education, determine the data centre of digital ideological and political education, and determine the dissimilarity of data through Minkowski distance; Using a random mapping method to convert the data into a consistent pattern, and implementing data preprocessing according to the optimal criteria for classification; calculate the aggregation weight of data, enable shared data to be back-propagated and iteratively updated, combine federated learning and incremental learning, build multiple data sharing clients, and add multiple data security keys to achieve secure sharing of digital ideological and political education data. The test results show that the method has good data sharing security and high sharing efficiency.
    Keywords: federal incremental learning; digital ideological and political teaching; data sharing; Minkowski distance; decision tree.
    DOI: 10.1504/IJRIS.2023.10058355
     
  • A Classification Method of Reader Borrowing Data Information in Modern Library Based on Top-k Query Algorithm   Order a copy of this article
    by Wei Huang, Jing Ling 
    Abstract: In order to overcome the problems of low accuracy of classification results and long classification time in the traditional classification method of modern library reader borrowing data information, a modern library reader borrowing data information classification method based on top-k query algorithm is proposed. First of all, top-k query algorithm is used to collect library readers’ borrowing data information and preprocess it; then, combining the information gain algorithm and the maximum correlation and minimum redundancy algorithm, the second feature selection is performed for the data information borrowed by readers; finally, the polynomial naive Bayesian model is used to realise the classification of library readers’ borrowing data information. The experimental results show that the classification results using this method are accurate, the classification time is always within 11 s, the classification effect is good, and the application performance is good.
    Keywords: top-k query algorithm; modern library; readers borrow data; information classification; maximum correlation minimum redundancy algorithm; polynomial naive Bayesian model.
    DOI: 10.1504/IJRIS.2023.10058356
     
  • Automatic classification of library digital resources based on decision tree algorithm   Order a copy of this article
    by Liping Zhong 
    Abstract: Aiming at the problems of large classification error, low accuracy of feature extraction and long classification time in automatic classification of library digital resources, this paper designs an automatic classification method of library digital resources based on decision tree algorithm. First, determine the maximum and shortest distance between digital resource data and aggregate the collected library digital resources. Then, the feature pyramid network structure is introduced and the channel multiplication method is used to extract the features of library digital resources data. Finally, construct the library digital resource data tree, determine the trunk and the critical degree of the library digital resource branches through entropy calculation, prune the unimportant branches, set up the library digital resource classifier, and realise the final automatic classification research. The test results show that the proposed method can reduce the classification error, and the classification effect is good.
    Keywords: decision tree algorithm; library digital resources; automatic classification; semantic similarity; the shortest distance; channel multiplication.
    DOI: 10.1504/IJRIS.2023.10058357
     
  • Multi label attribute classification algorithm for big data based on core density estimation   Order a copy of this article
    by Jian Xie, Dan Chu 
    Abstract: Aiming at the problems of inaccurate core density estimation of tag attributes during the classification process, resulting in large attribute feature extraction errors, low classification accuracy, and poor data balance coefficient, a large data multi tag attribute classification algorithm based on core density estimation is designed. First, the machine learning algorithm is used to extract data attribute features, calculate their expected estimates, and complete feature extraction. Then, the extracted data is processed in a unified dimension, and the mean value processing and Pearson correlation coefficient are used to complete the pre-processing. Finally, according to the core density estimation theory, determine the density of any data in the overall density function, then determine the probability quality function and degradation increment of other characteristic data points, introduce the classification kernel function, and construct the classification model to achieve the final effective classification. The results show that the proposed method can reduce feature extraction error and improve classification accuracy.
    Keywords: core density estimation; big data; multi-label classification; Pearson correlation coefficient; density function; degenerate increment.
    DOI: 10.1504/IJRIS.2023.10058358
     
  • Solving shortest path using Modified Dijkstra's algorithm with spherical neutrosophic numbers as arc length   Order a copy of this article
    by Prasanta Kumar Raut, Siva Prasad Behera 
    Abstract: The shortest path problem is a fundamental problem in network analysis, and many algorithms have been developed to solve it efficiently. One of the most widely used algorithms is Dijkstra’s algorithm, which finds the shortest path between two nodes in a graph. Recently, spherical neutrosophic numbers (SNNs) have emerged as a powerful tool for handling uncertain information in network domains. In this study, we propose a modified version of Dijkstra’s algorithm that uses SNNs to handle uncertainty in edge weights in a network. The edge weights are represented as SNNs to account for uncertainty in the distances between vertices, and the algorithm is modified to handle the arithmetic operations of SNNs and update distances and priorities accordingly. Our results show that the algorithm can effectively handle uncertainty in the network domain and find the shortest path between two nodes with high accuracy.
    Keywords: connected network; Dijkstra’s algorithm; shortest path problem; spherical neutrosophic numbers; SNNs.
    DOI: 10.1504/IJRIS.2023.10058359
     
  • Adaptive Classification of Electronic Music Signals Based on Multiple Machine Learning Models   Order a copy of this article
    by Shuqing Li 
    Abstract: In order to overcome the problems of low PSNR, low accuracy, and long time consumption in traditional methods, an adaptive classification method of electronic music signals based on multiple machine learning models is designed. Electronic music signals are collected by sensors and filtered by sparse coding. The filtered signal is processed by frame segmentation and windowing, and the time domain and frequency domain characteristics of the signal are extracted. The multi-machine learning model is built by Naive Bayes, support vector machine and decision tree. The signal filtering results are taken as the input vector of the model, and the adaptive classification results of electronic music signals are taken as the output vector of the model to realise the adaptive classification of signals. Experimental results show that the maximum PSNR of this method is 55.66 dB, the classification accuracy is always above 96%, and the average classification time is 867 ms.
    Keywords: multiple machine learning models; electronic music signals; adaptive classification; signal filtering; Naive Bayes; support vector machine; decision tree.
    DOI: 10.1504/IJRIS.2023.10058360
     
  • Personalized recommendation method of online education learning resources based on collaborative filtering algorithm   Order a copy of this article
    by Yang Yang 
    Abstract: In this paper, a personalised recommendation method for online education learning resources based on collaborative filtering algorithm is proposed. Firstly, the traditional crawler technology is used to collect resource data, and the implicit crawler technology is introduced to obtain key resources. Then, the automatic encoder extracts the nonlinear features, combines the expected risk minimisation and sample error, defines the empirical risk classification resource data, and completes the pre-treatment of resource data. Finally, a collaborative filtering algorithm is introduced to obtain learners’ individual needs through forgetting factor optimisation, and a collaborative filtering recommendation model is designed. The results show that the recommended error of the proposed method is 0.10%, the recommended time is less than two seconds, and the work efficiency is over 95%, indicating that the method can improve the accuracy and efficiency of resource recommendation.
    Keywords: collaborative filtering algorithm; online education; learning resources; personalised recommendation.
    DOI: 10.1504/IJRIS.2023.10058361
     
  • Recognition of Foul Actions of Football Players Based on Self Attention Mechanism   Order a copy of this article
    by Yingdong Song 
    Abstract: In order to solve the problems of low recognition accuracy and long recognition time in traditional methods, a recognition method of foul actions of football players based on self attention mechanism was designed. Firstly, the image of football player is obtained by Kinect device and the image features are extracted. Secondly, the background of action image is separated by background clipping method, and the noise of motion pixel point eight neighbourhood is removed by smoothing noise reduction method. Finally, set a set of key value pairs for foul actions, encode them using the encoder in the self attention mechanism, calculate the probability of foul actions occurring, and decode the dependency relationship of the encoded foul actions to determine whether the action is a foul action and obtain relevant recognition results. The test results show that the proposed method can improve the accuracy of foul action recognition and has a good effect.
    Keywords: self attention mechanism; football players; foul actions; recognition; background clipping method; encoder.
    DOI: 10.1504/IJRIS.2023.10058362
     
  • Comparison of Bald Eagle Search (BES) algorithm with benchmark meta-heuristic algorithms (PSO, ABC and GWO) applied to robot path planning   Order a copy of this article
    by Ganapati Kamat, Yogeswarr S, Narahari N, Vineeth Parashivamurthy 
    Abstract: This work focuses on comparing the latest swarm-intelligence-based algorithm, the bald eagle search (BES), with three well-known metaheuristic algorithms
    Keywords: bald-eagle search; BES; particle swarm optimisation; PSO; grey wolf optimisation; GWO; artificial bee colony; ABC; swarm intelligence; path optimisation.
    DOI: 10.1504/IJRIS.2023.10058592
     
  • Forest fire identification method of UAV remote sensing image based on FCM clustering algorithm   Order a copy of this article
    by Peiran Li, Yuqing Tan, Wei He, Haifeng Zhang, Zhanlan Xie 
    Abstract: In order to overcome the problems of low recognition accuracy and speed in traditional forest fire recognition methods, the paper proposes a forest fire recognition method based on unmanned aerial vehicle remote sensing images using FCM clustering algorithm. Firstly, the FCM clustering algorithm is used to cluster and segment the target RGB pixels in unmanned aerial vehicle remote sensing images. Secondly, according to the calculation rules of the LPB algorithm, the flame characteristics of forest fires are calculated. Finally, the optimal hyperplane of SVM is used to judge whether the target RGB pixels in the remote sensing image are fire pixels, and the forest fire recognition method can be obtained after traversing all pixels. The experimental results show that the fire location identified by this method is completely consistent with the actual situation, and the recognition rate can reach a maximum of 52 frames/s.
    Keywords: FCM clustering algorithm; UAV remote sensing image; forest fire awareness; flame characteristics.
    DOI: 10.1504/IJRIS.2023.10058888
     
  • Research on Korean translation error text detection method based on machine vision   Order a copy of this article
    by Ziyou Zhou 
    Abstract: Due to the differences and complexity between languages,Korean machine translation still has translation errors,ambiguity and discontinuity, such as translation errors, ambiguity and discontinuity.Therefore, this paper proposes a Korean translation error text detection method based on machine vision.The linear array CCD sensor in machine vision is used to collect Korean translation text images. The images are corrected for distortion and processed for grayscale,and enhanced through Gabor filters. The images are segmented into multiple candidate frames for recognizing the Korean translation text regions.Combining the results of text region recognition,a CNN-Attention model is constructed. The model is then used to input the image to be recognized, extract text features and match with knowledge points to output detection results. The experimental results show that the minimum text recognition rate of this method is 94.8%, the average detection rate is 97.1%, and the minimum detection time is 0.3s
    Keywords: Machine vision; Korean translation; Error text detection; Linear array CCD sensor; Gabor filters; CNN-Attention model.
    DOI: 10.1504/IJRIS.2023.10058890
     
  • Enterprise hidden financial information extraction method based on data source dependency   Order a copy of this article
    by Jingyi Li 
    Abstract: The significance of hidden financial information extraction research lies in the discovery of corporate financial fraud. In order to address the shortcomings of traditional methods such as low recall and precision, and long time overhead, therefore an enterprise hidden financial information extraction method based on data source dependency is proposed. By determining the redundancy of enterprise financial information through data source dependency relationships, the data source dependency relationships are cleaned, and the cleaned data is input into the RDCNN-CRF model to achieve information label classification. Combined with the label classification results, an ordered long-short-term memory-multi-head attention mechanism neural network model is constructed, and the processed data is input into this model. The model output is the result of extracting hidden financial information. The experimental results show that the mean recall rate and precision rate of the proposed method are 97.92% and 98.07%, and the maximum time consumption is 0.82 s.
    Keywords: data source dependency; enterprise; hidden financial information; information extraction; redundancy; RDCNN-CRF model.
    DOI: 10.1504/IJRIS.2024.10059318
     
  • A Personalized Recommendation Method for Low-carbon Tourism Routes Based on User Feature Mining   Order a copy of this article
    by Bing Hou 
    Abstract: In order to overcome the problem of poor user feature mining accuracy and recommendation satisfaction in personalised recommendation methods, a low-carbon tourism route personalised recommendation method based on user feature mining is proposed. Firstly, determine the changes in the cost of low-carbon tourism production behaviour and the demand curve for low-carbon tourism, as well as the characteristics of low-carbon tourism. Secondly, calculate the similarity of low-carbon tourism user features to achieve overall feature mining for low-carbon tourism users. Finally, calculate the user feature entropy and conditional entropy, and analyse the key degree of low-carbon behaviour characteristics of tourism users. Personalise recommendations for low-carbon tourism routes that meet the conditions based on users’ preferred low-carbon tourism attractions. The experimental results indicate that the research method can effectively improve the efficiency of fine-grained user feature mining, and the satisfaction with personalised recommendation of low-carbon tourism routes is high.
    Keywords: user feature mining; low carbon tourism; route planning; personalised recommendation; conditional entropy; route planning.
    DOI: 10.1504/IJRIS.2024.10059319
     
  • An optimizing method of Japanese character recognition based on improved support vector machine   Order a copy of this article
    by Yang Wang, Pan Zhao 
    Abstract: Aiming at the problem of low recognition accuracy and long recognition time due to poor feature extraction, a Japanese character recognition method based on improved support vector machine was studied. First, the mean filtering method is used to denoise Japanese text samples, and then Gabor filter is used to extract features, and Fourier transform is used to optimise the extraction speed. Then, the initial text recognition method is constructed by introducing support vector machine and improving the penalty coefficient algorithm. Finally, Japanese character recognition is realised by using the multi-layer perceptron function as the kernel by processing linear indivisible samples with the relaxation term minimisation. The experimental results show that the proposed method can effectively extract Japanese character features with a recognition accuracy of 97.4% and a recognition time of only 1.3 s, which effectively solves the problems of low recognition accuracy and long recognition time.
    Keywords: support vector machine; Japanese characters; identification method; Gabor filter; Fourier transform; penalty coefficient; Kernel function.
    DOI: 10.1504/IJRIS.2024.10059320
     
  • Evaluation of the shortest path using a ripple-spreading algorithm in an uncertain environment   Order a copy of this article
    by Prasanta Kumar Raut, Siva Prasad Behera 
    Abstract: The efficient determination of the SP in a connected network is a crucial problem in various domains, such as transportation, communication, and logistics. Traditional approaches often assume crisp values for the parameters, but in real-world scenarios, uncertainties and imprecision are prevalent. This study proposes a novel approach that utilises the ripple-spreading algorithm with triangular fuzzy numbers as parameters to determine the SP in a given network. The ripple-spreading algorithm, known for its ability to propagate information or updates through a network, is adapted to handle uncertainties by incorporating triangular fuzzy numbers. Each connection in the network is described by a triangular fuzzy number, which utilises the triangular fuzzy numbers consisting of lower bounds, modal values, and upper bounds, forming a triangular-shaped membership function. The propagation is performed considering the triangular fuzzy numbers as parameters, accounting for uncertainties and imprecisions.
    Keywords: ripple-spreading algorithm; shortest path; network; triangular fuzzy number; uncertainty.
    DOI: 10.1504/IJRIS.2024.10059321
     
  • Study on Deep Mining of Network Ideological and Political Resources Based on Weighted Deep Forest   Order a copy of this article
    by Dan Yin, Na Wang 
    Abstract: Traditional methods for deep mining of ideological and political resources on the internet have problems such as poor comprehensiveness in mining and ineffective fine-grained control. Therefore, a deep mining method of network ideological and political resources based on weighted deep forest is proposed. Firstly, the ideological and political resources are classified and their relationships are structured into a tree-like structure. Then, keyword tables, index tables, and synonym index tables are designed to extract features of resources after indexing and dimensionality reduction. Finally, a multi-granularity scanning structure based on weighted deep forest is constructed to determine the existence status of resources on different network platforms, and resource deep mining is achieved by determining data weight factors. By analysing the experimental results, it can be seen that the method proposed in this article has good comprehensiveness in deep resource extraction, good fine-grained control effect, and good mining effect.
    Keywords: weighted deep forest; network ideological and political resources; deep mining; tree-like structure; extract features; weight factors.
    DOI: 10.1504/IJRIS.2024.10059322
     
  • Adverse COVID-19 Vaccination-Related Events in India: A Cross-sectional Study Using Machine Learning to Predict Their Severity   Order a copy of this article
    by Hemangini Mohanty, Santilata Champati, Jyotiranjan Sahoo 
    Abstract: The most effective method of preventing Coronavirus illness is vaccination, despite its status as a global outbreak. In this study, India's COVID-19 vaccination's unfavorable consequences are evaluated and builds a model for predicting the severity of side effects. A cross-sectional study was conducted through an online survey among the Indian population who received at least a single dose of COVID-19 vaccine. Then, data were statistically analyzed, and machine learning tools were used to build a predictive model predicting the severity of side effects. A total of 3222 participants' records were analyzed; for those participants receiving three different Vaccines, i.e., Covaxin, Covishield, and SputnikV. Only 25% experienced mild-to-moderate side effects. The most common side effects recorded were Fever/Chills, headache, feeling pain at the injection site, tiredness and fatigue. The people receiving the first dose (71.93%) had significant side effects compared to the second dose.
    Keywords: SARS-CoV-2 vaccine; COVID-19 vaccine; post-vaccination symptoms; machine learning; predictive model.
    DOI: 10.1504/IJRIS.2024.10059323
     
  • Content Clustering Based Propagation Feature Extraction for Short Video Media Platforms   Order a copy of this article
    by Shiming Zhou 
    Abstract: In order to improve the accuracy of short video media platform propagation feature extraction, this paper proposes a content clustering based propagation feature extraction method for short video media platforms. Firstly, the data obtained through data crawling is user online comments, which are used to obtain user comments on short video media platforms; Secondly, preprocess the propagation data of short video media platforms through data cleaning and text segmentation; Then, input the short video, calculate the classification loss for each frame separately and sum it up. Finally, the content clustering method is used to cluster the propagation features of short video media platforms, and the final propagation features of short video media platforms are obtained by solving the propagation feature function. The experimental results show that the proposed method can effectively improve the accuracy of propagation feature extraction and enhance the recall rate of feature extraction.
    Keywords: K-means clustering; Content clustering; Data crawling; Cross entropy loss.
    DOI: 10.1504/IJRIS.2023.10059840
     
  • Personalized Recommendation Method of Japanese Teaching Resources Based on Collaborative Filtering   Order a copy of this article
    by Pan Zhao, Yang Wang 
    Abstract: In the existing personalized recommendation methods of Japanese teaching resources, the confidence of resource recommendation is low, and the personalized recommendation effect is poor. Therefore, a personalized recommendation method for Japanese teaching resources based on collaborative filtering is designed. Firstly, the multi-modal interest similarity model and hidden factor depth are used to extract the features of Japanese teaching resources. Then, the scoring matrix and label matrix are constructed, and TF-IDF algorithm is introduced to calculate user preferences, so as to realize user preference mining of Japanese teaching resources. Finally, the Japanese teaching resources recommendation scoring matrix is constructed, similar neighbors with preferences are sought, and personalized recommendation rules are set to realize personalized recommendation of Japanese teaching resources. The experimental results show that the proposed method can improve the confidence of Japanese teaching resource recommendation and improve the poor performance of personalized recommendation.
    Keywords: Collaborative filtering; Japanese language teaching resources; Preference; Scoring matrix; Label matrix; Penalty factor.
    DOI: 10.1504/IJRIS.2023.10059841
     
  • An online teaching resource recommendation algorithm based on category similarity   Order a copy of this article
    by Lingyu Chen 
    Abstract: To overcome the problems of poor recall rate, time-consuming resource recommendation list generation, and low ideal loss cumulative gain in traditional online teaching resource recommendation algorithms, a category similarity based online teaching resource recommendation algorithm is proposed. Firstly, segment the user group of online teaching resources and construct a learning user profile of online teaching resources based on dynamic field theory; Secondly, the similarity between user categories and teaching categories is obtained, and the similarity of online teaching resource categories is obtained through sorting; Finally, based on category similarity, an online teaching resource PAF recommender is constructed, taking into account teaching popularity and user loyalty to achieve online teaching resource recommendation. The experimental results show that the online teaching resource recommendation recall rate of this algorithm can reach 99.9%, the ideal cumulative loss gain of teaching resource recommendation is 66.18, and the resource recommendation list generation time is 11 s.
    Keywords: category similarity; teaching resources; recommendation algorithm; dynamics; product adoption forecasting; PAF recommender.
    DOI: 10.1504/IJRIS.2023.10059842
     
  • Network Intrusion Detection in Smart Grid using Supervised Learning   Order a copy of this article
    by Jason Marandi, Priyanka Ahlawat 
    Abstract: The emergence of smart grid (SG) as a replacement for the traditional power grid has increased the efficiency and reliability of the power grids. This is due to incorporating communication networks with the traditional power grid. This integration also exposes the grid to vulnerabilities common in communication networks. An intrusion detection system (IDS) is an important tool for providing secure and reliable services in smart grid. Anomaly-based IDS (AIDS) has gained importance for detecting zero-day attacks. Machine learning (ML) is a promising solution as it learns from past events to predict new network threats. Two common ML types are supervised and unsupervised learning. They differ based on how data is presented to them. This paper explores supervised ML approaches using the CICIDS2017 dataset to achieve the best classifier and parameters. The tuned random forest (RF) model achieved an impressive 99.92% accuracy with a 70:30 dataset split for training and testing.
    Keywords: smart grid; SG; intrusion detection system; IDS; signature based detection; anomaly based detection; IDS datasets; machine learning; ML; supervised learning.
    DOI: 10.1504/IJRIS.2023.10060968
     
  • Multi channel user interface generation method based on conflict degree and collaborative filtering   Order a copy of this article
    by Ting Zhang 
    Abstract: Aiming at the problems of low recommendation accuracy, low success rate of interactive control and long time of traditional recommendation methods, a multi-channel user interface generation method based on conflict degree and collaborative filtering is proposed. Firstly, the evidence optimality and binary comparison matrix of user behaviour are determined, and the evidence conflict degree of user behaviour is calculated using the weight calculation results, and user behaviour recognition is realised according to the evidence characteristics. Secondly, collaborative filtering is adopted to implement user interface pattern recommendation. Finally, according to the recommended user interface model, a multi-channel user interface generation framework is constructed, including user task decomposition, multi-channel interaction control and interface code determination. Experimental test results show that the maximum accuracy of user interface pattern recommendation using proposed method is 98.36%, the average success rate of multi-channel interaction control is 97.35%, and the minimum time for multi-channel user interface generation is 1.8 s.
    Keywords: conflict degree; collaborative filtering; multi-channel; user interface generation; evidence optimality; multi-channel interaction control.
    DOI: 10.1504/IJRIS.2024.10061315
     
  • An Abnormal Detection method of Enterprise Financial Accounting Data Based on Bayesian network   Order a copy of this article
    by Baoyuan Liu 
    Abstract: To improve the effectiveness of anomaly detection in enterprise financial accounting data and reduce the error probability of anomaly detection, this paper proposes a Bayesian network-based anomaly detection method for enterprise financial accounting data to ensure the accuracy and reliability of financial reports. By introducing the nearest neighbour rule and KNN algorithm to calculate the distance between different data attributes, the XGBoot algorithm is used to obtain the optimal balance point and achieve the classification of enterprise financial accounting data; According to the topological structure, the prior knowledge and accounting data characteristics are fitted, the accounting data characteristics are extracted, the interference items of abnormal features are removed by Markov blanket elimination method, the conditional probability of Bayesian network is calculated, and the data anomaly detection is realised to realise the final research. The test results indicate that the false positive rate of abnormal data detected by this method is low, and the recall rate is high, which has certain feasibility.
    Keywords: Bayesian network; enterprise financial accounting data; abnormal detection; nearest neighbour rule; parameter learning.
    DOI: 10.1504/IJRIS.2024.10061316
     
  • A Modified Algorithm to Solve Minimum Spanning Tree Problem   Order a copy of this article
    by Siva Behera, Prasanta Kumar Raut 
    Abstract: Here, we give a modified and improved algorithm to find an MST from a given weighted undirected graph. Our algorithm works in two stages. In the first stage, it finds a forest formation involving the edges with minimum weights, and in the second stage, the algorithm converts the forest obtained in the first stage into a spanning tree. Further, we demonstrate our algorithm with one example, discuss the complexity of our algorithm, give its implementation using the Java Applet program, and compare the output with Kruskal’s algorithm. This study demonstrates that the suggested approach is superior to conventional algorithms for handling MST.
    Keywords: array; binary heap; Kruskal’s algorithm; minimum spanning tree; forest; stack.
    DOI: 10.1504/IJRIS.2024.10061376
     
  • Color Offset Compensation Method of 3D Animation Scene Image Based on Color Difference Interpolation   Order a copy of this article
    by Xu Lan, Lizhen Jiang 
    Abstract: A colour offset compensation method for 3D animation scene images based on colour difference interpolation is proposed to address the issues of unsatisfactory colour offset compensation and long time in traditional 3D animation scene images. Adopting high-dimensional convolutional fast bilateral filtering algorithm for denoising 3D animation scene images, using an improved gain coefficient greyscale world correction method for correction, converting the image from RGB space to HIS colour space, linearly stretching the saturation component of the image, and achieving colour offset compensation for 3D animation scene images. The simulation results show that after applying the method proposed in this paper for image processing, the highest signal-to-noise ratio is 23.85db, the highest peak signal-to-noise ratio is 28.69db, the denoising running time is always kept below 125ms, the maximum information entropy is 8.30, the maximum contrast is 72.80, the minimum colour difference is 0.521, and the colour offset compensation time is always kept below 90ms. The colour offset compensation effect is good.
    Keywords: colour difference interpolation; 3D animation scene image; colour offset compensation.
    DOI: 10.1504/IJRIS.2024.10061377
     
  • Speech recognition method of English translation robot based on HMM algorithm   Order a copy of this article
    by Xiaolin Zhang, Tao Wang, Ling Jiang 
    Abstract: To solve the problems of poor speech feature extraction and high false recognition rate in English translation robot speech recognition, a speech recognition method for English translation robot based on HMM algorithm is proposed. Firstly, the speech signal is collected and processed by noise reduction, reverberation removal, speech segmentation, volume normalisation and speech feature extraction. Then, HMM algorithm is introduced to extract the optimal speech features. Finally, the speech recognition function is constructed, and the speech features are regarded as discrepancy, and the recognition function is optimised through the error term to realise the speech recognition of the English translation robot. The experimental results show that this method can effectively extract the voice features of voice command signals. The average error rate is 1.5%, and the average recognition rate is 98.47%. The identification results of valid information and invalid information obtained are consistent with the actual results.
    Keywords: HMM algorithm; English translation robot; speech recognition; noise reduction; de-reverberation; speech segmentation; volume normalisation; speech feature extraction.
    DOI: 10.1504/IJRIS.2024.10061378
     
  • A Multi object Tracking Method for Complex Scenes Based on Edge Feature Extraction of Video Images   Order a copy of this article
    by Haidi Yuan, Wei Li 
    Abstract: In order to improve the accuracy of complex scene tracking and shorten tracking time, this paper proposes a multi object tracking method for complex scenes based on video image edge feature extraction. Firstly, capture video images from complex scenes. Secondly, obtain the edge feature components of video images and extract complex edge features of video images. Then, a single target tracker is designed to construct an affinity matrix and filter out elements that exceed the threshold. Finally, an undirected graph is constructed to match the previously tracked trajectory with the detection results. For trajectory updates on the matching, multi target tracking is achieved by solving the maximum weight clique graph to achieve complex scene multi target tracking. The experimental results show that the accuracy of our method can reach 99.89%, and the tracking time is only 2.6 seconds, indicating that our method can effectively improve the tracking effect.
    Keywords: template matching; sports affinity; maximum weight clique graph; affinity function.
    DOI: 10.1504/IJRIS.2024.10061379
     
  • Intelligent Management Platform for Enterprise Dynamic Accounting Information Based on Data Mining   Order a copy of this article
    by Huiying Kang 
    Abstract: In order to promote the improvement of enterprise financial management level, a data mining-based intelligent management platform for enterprise dynamic accounting information is proposed. Firstly, the SOM algorithm is used to fully collect dynamic accounting information of enterprises, and the collected data is used as the management object. Secondly, the association rule method is used to calculate the support and confidence levels, completing the normalisation of accounting information. Finally, based on the normalised accounting information, a quantitative management function for accounting information is constructed to achieve intelligent management of accounting information. The experimental results show that the average accounting information mining time of the platform in this article is around 8 minutes, the maximum access latency is not more than 10 ms, and the accounting information reading time is always less than 2.5 s. The performance of the intelligent management platform has been significantly improved.
    Keywords: data mining; dynamic accounting information; intelligent management platform; SOM algorithm; normalisation processing.
    DOI: 10.1504/IJRIS.2024.10061380
     
  • Enterprise financial anomaly data detection method based on improved support vector machine   Order a copy of this article
    by Hao Wang, Huan Wang 
    Abstract: In this paper, a method for detecting financial anomaly data in enterprise based on improved support vector machine is proposed. By analyzing the abnormal financial data of enterprises, the distribution of abnormal financial data is determined, and a multi-channel Text-CNN neural network model is constructed to extract initial abnormal data features. The nonlinear features of abnormal data are adjusted using the least squares method to achieve feature extraction of abnormal financial data of enterprises. Optimize the support vector machine algorithm through differential evolution algorithm to determine the optimal classification population for enterprise financial anomaly data features for global optimization; By constructing a global selection model, achieve the detection of abnormal financial data in enterprises. The test results indicate that the fitting degree of the method proposed in this paper is good, and the detection error is low, indicating a certain degree of feasibility.
    Keywords: Improving support vector machines; Point anomaly; Context exception; Collection anomaly; Least squares method; Differential Evolution Algorithm.
    DOI: 10.1504/IJRIS.2024.10061381
     
  • Latest advances in deep learning based recommender systems   Order a copy of this article
    by Amina Debbah, Samira Lagrini 
    Abstract: Recommender systems (RSs) are prominent tools massively used in different fields of social life, e-commerce, and online platforms. The use of machine learning techniques to build RSs gives good results, but it cannot satisfy all user’s requirements nowadays. The exponential growth of available data integrating social networks, and contextual and temporal properties, greatly complicate the control of user’s request using traditional machine learning and decision support systems, as these techniques fail to handle massive multimedia data sources, and cannot effectively capture nonlinear relationships between users and items. Moreover, they dismiss context, social relationships, and trustworthiness. Currently, deep learning techniques are successfully used in almost artificial intelligence fields including RSs. Latest researches proves that deep learning based RSs yield promising results and outperforms traditional machine learning techniques. This paper provides a comprehensive overview of recent advances in deep learning based RSs. We deeply analyse the challenges of these systems, and how recent research works address these challenges. Furthermore, we address the strengths and weaknesses of existing approaches in order to offer an exhaustive view for new researchers to develop new ideas when tackling the issue of RSs.
    Keywords: recommender systems; RSs; deep learning; collaborative filtering; content filtering; hybrid filtering; neural networks; evaluation metrics.
    DOI: 10.1504/IJRIS.2024.10062538
     
  • Anomaly detection for dual-channel sleep EEG signal with Mahalanobis-Taguchi-Gram-Schmidt metric   Order a copy of this article
    by Jiufu Liu, Rui Zheng, Zaihong ZHOU, Zhong Yang, Zhisheng Wang 
    Abstract: To realise the automatic and accurate detection of human sleep quality, to overcome the complex and complicated process caused by artificial subjective discrimination, this paper presents a measurement algorithm of sleep EEG signals based on Mahalanobis-Taguchi-Gram-Schmidt model. The characteristic vectors of each channel are normalised in different staging segments, and the Schmidt orthogonal vector Group of the linear independent vectors is obtained. The signal-to-noise ratio mean value of EEG in six-state sleep stages, in 30-minute periods and 15-minute periods is calculated respectively with Mahalanobis-Taguchi-Gram-Schmidt method. The measurements of six-state sleep stages are analysed to identify and determine the normal and anomaly sleep quality. Mahalanobis-Taguchi-Gram-Schmidt metric for sleep EEG signals is effective in the detection of human sleep quality.
    Keywords: anomaly detection; Mahalanobis-Taguchi system; sleep stages; signal to noise ratio.
    DOI: 10.1504/IJRIS.2024.10062539
     
  • Fog Computing Approaches for Sustainable Smart Cities   Order a copy of this article
    by Parimal Giri, Sanjoy Choudhury, Diptendu Sinha Roy, Bijay Paikaray 
    Abstract: Reverse auction resource allocation and utilisation in a fog environment is an exciting yet challenging problem due to the numerous constraints and needs. The end user presents the resource needed and exposes it to a notable set of qualified expert suppliers in a reverse auction mechanism. This paper provides a combinatorial reserve auction for the fog environment that considers both cost and non-cost attributes, such as the type of QoS criteria, reputation, and other factors, to ensure the success of winning stakeholder cooperatives. To solve this problem, estimate calculation is used, and a polynomial-time solution that comes near to being exact is obtained. In order to increase their benefit, the auction process allows suppliers to disclose accurate data. This strengthens the system since it can keep up with the client’s utility. According to the execution assessment and a relative report utilising various widely used models, the suggested method performs better.
    Keywords: fog computing; CRA; smart cities.
    DOI: 10.1504/IJRIS.2024.10062540
     
  • AMAA-GMM: Adaptive Mexican Axolotl Algorithm based Enhanced Gaussian Mixture Model to Segment the Cervigram Images   Order a copy of this article
    by Lalasa Mukku, Jyothi Thomas 
    Abstract: Colposcopy is a crucial imaging technique for finding cervical abnormalities. Colposcopic image evaluation, particularly the accurate delineation of the cervix region, has considerable medical significance. Before segmenting the cervical region, specular reflection removal is an efficient approach. Because, cervical cancer can be found using a visual check with acetic acid, that turns precancerous and cancerous areas white and these could be viewed as signs of abnormalities. Similarly, bright white regions known as specular reflections obstruct the identification of aceto-white areas and should therefore be removed. So, in this paper, specular reflection removal with segmenting the cervix region in a colposcopy image is proposed. The proposed approach consists of two main stages, namely, pre-processing and segmentation. In the pre-processing stage, specular reflections are detected and removed using a swin transformer. After that, cervical regions are segmented using an enhanced Gaussian mixture model (EGMM). For better segmentation accuracy, the best parameters of GMM are chosen via the adaptive Mexican axolotl optimisation (AMAO) algorithm. The performance of the proposed approach is analysed based on accuracy, sensitivity, specificity, Jaccard index, and dice coefficient, and the efficiency of the suggested strategy is compared with various methods.
    Keywords: Gaussian mixture models; machine learning; segmentation; metaheuristics; deep learning; enhanced Gaussian mixture model; EGMM; adaptive Mexican axolotl optimisation; AMAO.
    DOI: 10.1504/IJRIS.2024.10063302
     
  • Modified VGG19 Transfer Learning Model for Breast Cancer Classification   Order a copy of this article
    by Sashikanta Prusty, Srikanta Patnaik, Sujit Kumar Dash 
    Abstract: Breast cancer (BC) seems to have become a sign of great concern in everyday life. There have been a lot of research and methods already designed in the last few years but continue to be prone worldwide. To address this issue, a modified version of the visual geometric group-19 (VGG19) model, namely BCNet21 has been proposed here to classify the malignant class from breast mammogram images collected from the MIAS dataset. Furthermore, the performance of our proposed BCNet21 model has been compared with the two most common predefined VGG16 and VGG19 models using the performance metrics and Cohen-Kappa test (k). The result shows that the proposed BCNet21 model outperforms with a higher accuracy of 98.96 % and a kappa score of 86%, compared to the VGG16 and VGG19 models. This concludes that the BCNet21 model is much closer to the near-perfect agreement between actual and predicted breast cancer instances.
    Keywords: breast cancer; BC; deep learning; DL; transfer learning; TL; VGG19; VGG16; kappa score.
    DOI: 10.1504/IJRIS.2024.10063303