Applied Soft Computing最新文献

{"title":"Knowledge-enhanced large language models for ideation to implementation: A new paradigm in product design","authors":"Zhinan Li ,&nbsp;Zhenyu Liu ,&nbsp;Guodong Sa ,&nbsp;Jiacheng Sun ,&nbsp;Mingjie Hou ,&nbsp;Jianrong Tan ,&nbsp;Lei Sun ,&nbsp;Jun Wei","doi":"10.1016/j.asoc.2025.113147","DOIUrl":"10.1016/j.asoc.2025.113147","url":null,"abstract":"<div><div>Traditional product design processes often struggle to accurately capture complex user needs and generate market-relevant solutions due to a heavy reliance on subjective human input and limited decision support tools. While Large Language Models (LLMs) have shown proficiency in various domains, their application in product design remains limited, often resulting in generic outputs. To address this, we propose an innovative paradigm for intelligent product design generation, termed ProdGen. The core of ProdGen is the ProdGen-Agent system, which integrates LLMs with customized expert design tools, leveraging the proposed Multi-Design Task Adapter (MDT-A) method and a Dual Knowledge Enhancement Mechanism. The MDT-A method injects multimodal design task knowledge into LLMs through a unified knowledge fusion framework, enabling enhanced task decomposition and efficient interaction with custom design tools. The Dual Knowledge Enhancement Mechanism enriches LLM performance by incorporating domain-specific knowledge bases and structured graph-based data retrieval, ensuring more accurate and relevant design outputs. Demonstrated through kitchen design cases, ProdGen-Agent autonomously handles the entire design process, excelling in user need analysis, task breakdown, decision-making support, tool integration, and multidimensional design generation. Expert evaluations validate ProdGen-Agent’s effectiveness in automating complex design tasks, confirming its potential to revolutionize product design processes across various industries by leveraging LLMs in combination with domain expertise.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"176 ","pages":"Article 113147"},"PeriodicalIF":7.2,"publicationDate":"2025-04-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143854813","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Multi-fault diagnosis with wavelet assisted stacked image fusion and dual branch CNN","authors":"Rismaya Kumar Mishra ,&nbsp;Anurag Choudhary ,&nbsp;S. Fatima ,&nbsp;A.R. Mohanty ,&nbsp;B.K. Panigrahi","doi":"10.1016/j.asoc.2025.113183","DOIUrl":"10.1016/j.asoc.2025.113183","url":null,"abstract":"<div><div>The rotating machine components are interconnected. If the machines are not monitored properly, it causes damage to the connected parts, causing catastrophic failure. Dependability on a single sensor or sensors of the same modality for multi-fault diagnosis influences decision-making. Therefore, multi-modality multi-sensor fusion has been used to gather distinct information. This work proposes a Wavelet Assisted Stacked Image Fusion (WASIF) with Dual Branch Convolutional Neural Network (DBCNN) to effectively diagnose multi-faults. At first, various multi-fault conditions in a test rig are introduced, which consist of conditions like faulty motor, faulty bearing, mechanical unbalance, shaft misalignment and their combinations. Thereafter, vibration and acoustic data are acquired at a varying speed condition. The acquired signatures are pre-processed and converted into time-frequency spectrums using Fourier Synchrosqueezed Transform (FSST). The vibration and acoustic spectrums are fused into vibro-acoustic spectrums using the WASIF technique. The generated spectrums are used for DBCNN training for multi-fault classification, and 98.8 % overall classification accuracy is achieved. In this paper, a separate ablation experiment is done along with a published literature comparison to justify the effectiveness of the selected parameters. The proposed fusion-based multi-fault diagnosis strategy would be helpful to the industries for incipient fault detection, inventory management and workforce allocation.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"176 ","pages":"Article 113183"},"PeriodicalIF":7.2,"publicationDate":"2025-04-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143864963","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Service defects identification by integrating fuzzy clustering and optimization model with quality function deployment","authors":"Xiaobing Li ,&nbsp;Yujun Wang ,&nbsp;Zhen He","doi":"10.1016/j.asoc.2025.113175","DOIUrl":"10.1016/j.asoc.2025.113175","url":null,"abstract":"<div><div>Service defect is critical for the success of fresh product instant delivery (FPID) service. Therefore, the identification and prioritization of such defects is an important demand from FPID companies. Quality function deployment (QFD) can help companies identify and prioritize service defects. However, existing research based on QFD failed to consider the subjectiveness and similarity of evaluation information simultaneously while neglecting the integration of service quality characteristics and customer expectations. To fill these gaps, a novel methodology based on QFD model and optimization model is proposed. Firstly, a two-phase QFD framework for FPID service is developed to establish service quality characteristics, strengthening the relationship between customer requirements and service quality characteristics. To address the fuzziness and similarity of evaluation information, a new analysis model, named L-FCM-HOQ, by integrating Linguistic terms, Fuzzy C-Means (FCM), and House of Quality (HOQ) is proposed to determine related weights in QFD. Secondly, a decision variable, \"improvement rate\" is introduced to determine service defects among service quality characteristics and quantify their improvement degrees. Based on this decision variable, a multi-objective optimization model is constructed to derive optimal improvement strategies aligned with customer expectations. Finally, the proposed methodology is illustrated with a case study regarding a FPID company of China, and its efficiency and advantages are verified via comparative analysis. The proposed methodology effectively identifies service defects and assesses improvement potential, contributing to the development of FPID service quality and actionable insights to guide the FPID companyies in achieving sustainable improvement.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"176 ","pages":"Article 113175"},"PeriodicalIF":7.2,"publicationDate":"2025-04-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143868642","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Semi-supervised feature contrast incremental learning framework for bearing fault diagnosis with limited labeled samples","authors":"Xuyang Tao ,&nbsp;Changqing Shen ,&nbsp;Lin Li ,&nbsp;Dong Wang ,&nbsp;Juanjuan Shi ,&nbsp;Zhongkui Zhu","doi":"10.1016/j.asoc.2025.113172","DOIUrl":"10.1016/j.asoc.2025.113172","url":null,"abstract":"<div><div>In real-world scenarios, rotating machinery consistently introduces new fault classes, but intelligent fault diagnosis methods mostly rely on the closed-world assumption, expecting only known fault classes during testing. Moreover, obtaining a sufficient number of labeled samples is often challenging. These challenges constrain the application and reliability of intelligent diagnosis models in real-world scenarios. Semi-supervised incremental learning enables continuous learning of new fault classes in an open environment, relying on a small number of labeled samples and a certain number of unlabeled samples. To address the semi-supervised incremental learning problem of fault classes, semi-supervised feature contrast (SSFC) is proposed, a new approach for bearing fault diagnosis with limited labeled samples. Specifically, a feature contrastive loss incorporating enhancement strategies is designed, independent of labeled sample information. This approach enables the model to retain knowledge of old classes while learning about new ones. A label reconstruction mechanism based on class centroids is utilized, effectively leveraging the structural information inherent in the samples to support supervised training. A dynamic class prototype cosine classifier initialized by class centroids is devised to mitigate interference between knowledge of fault classes. Finally, two incremental fault diagnosis case studies are designed to evaluate the effectiveness of the proposed method. The fault diagnosis results indicate that SSFC can continuously learn knowledge of new fault classes with limited labeled samples and effectively alleviate catastrophic forgetting.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"176 ","pages":"Article 113172"},"PeriodicalIF":7.2,"publicationDate":"2025-04-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143855479","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Wavelet-denoised graph-Informer for accurate and stable wind speed prediction","authors":"Biao Yu,&nbsp;Zhenyu Lu,&nbsp;Weiwei Qian","doi":"10.1016/j.asoc.2025.113182","DOIUrl":"10.1016/j.asoc.2025.113182","url":null,"abstract":"<div><div>As global demand for renewable energy increases, wind power has become increasingly valued as a clean energy source. Effective wind speed forecasting is crucial for wind energy production and power grid's stability. To mitigate high-frequency interference in wind speed signals and improve spatiotemporal feature extraction, we propose a novel short-term wind speed prediction model called WST-Informer. Firstly, Wavelet Decomposition (WD) is fused to erase high-frequency noise in monitored signal series. Secondly, Informer encoder is designed to capture long-term temporal dependencies efficiently, and multiple cities' spatio-temporal maps are constructed through designing Residual Graph Convolutional Network (RS-GCN). Moreover, a new Attentional Feature Fusion (AFF) method is designed to fuse temporal and spatial features. Furthermore, the decoder of the Informer predicts outcomes using fused features. Additionally, outliers are more prone to bigger errors and highly curtail in real-world application, a Kernel Mean Squared Error (KMSE) loss function is introduced to further enhance their prediction. With real datasets from weather stations across five Danish cities and seven Dutch cities, extensive experiments were conducted, and the proposed model demonstrates reduced prediction error across multiple forecasting steps in both datasets, resulting in lower prediction errors during sudden wind speed changes, outperforming current state-of-the-art time series forecasting models.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"176 ","pages":"Article 113182"},"PeriodicalIF":7.2,"publicationDate":"2025-04-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143848445","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"BORF: A Bayesian optimized random forest for prediction of aerosol extinction coefficient from Mie Lidar signal","authors":"Hao Chen ,&nbsp;Fei Gao ,&nbsp;Zhimin Rao ,&nbsp;Dengxin Hua","doi":"10.1016/j.asoc.2025.113130","DOIUrl":"10.1016/j.asoc.2025.113130","url":null,"abstract":"<div><div>In continuous observation signals of lidar, the identification and selection of effective signals are crucial, especially for the aerosol extinction coefficient retrieval. In this study, the Bayesian Optimized Random Forest (BORF) model, a machine learning approach combining Random Forest regression with Bayesian optimization, was developed for predicting aerosol extinction coefficients. Built upon the foundation of the Random Forest (RF) regression method, this model leverages Bayesian optimization to adjust model parameters precisely, significantly enhancing the accuracy of aerosol extinction coefficient predictions. This approach offers a valuable means to identify and screen anomalous Lidar signals. We constructed a training dataset comprising continuously observed Mie Lidar signals and aerosol extinction coefficients retrieved using the Klett method. The dataset contains dimensions, including Mie Lidar signals, detection time, detection distance, pressure, and temperature. This paper provides a detailed description of the BORF model’s establishment process and the optimization of model parameters using Bayesian optimization. Through model assessments, significance tests, and comparative experiments, we demonstrate the effectiveness of the BORF model. Experimental results indicate that, compared to other relevant models, the BORF model excels in predicting aerosol extinction coefficients, closely aligning with the accuracy of the Klett method. Specifically, in datasets with better data quality, the BORF model exhibits an approximately 4% increase in <span><math><msup><mrow><mi>R</mi></mrow><mrow><mn>2</mn></mrow></msup></math></span> compared to the RF and BP neural network optimized by genetic algorithm (BPGA), accompanied by a 41% to 47% reduction in MSE and MAE. The Mean Squared Error (MSE) and Mean Absolute Error (MAE) decrease by approximately 40% to 90% in datasets with lower data quality and less apparent data variations. This study provides a robust technical solution to ensure the reliability of Lidar data, thereby contributing to an enhanced understanding of atmospheric aerosols and environmental monitoring.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"176 ","pages":"Article 113130"},"PeriodicalIF":7.2,"publicationDate":"2025-04-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143864964","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Constrained multi-objective optimization via neural network and cooperative populations","authors":"Jie Cao ,&nbsp;Yiyuan Wang ,&nbsp;Jianlin Zhang ,&nbsp;Zuohan Chen","doi":"10.1016/j.asoc.2025.113051","DOIUrl":"10.1016/j.asoc.2025.113051","url":null,"abstract":"<div><div>Constrained multi-objective optimization problems are widely used in practical scenarios such as intelligent manufacturing and network communication. These problems are often made intractable by constraints, and achieving a balance between convergence, diversity, and feasibility becomes increasingly challenging. To address this issue, a constrained multi-objective evolutionary algorithm named NNCP is proposed, which is based on the neural network and, three cooperative populations. Specifically, the neural network is employed to accelerate the population’s convergence by utilizing neuron weights to capture neighborhood information. Among the three populations, the first population uses self-organizing mapping and curvature estimation to approximate the Pareto front, the second population utilizes non-dominance sorting and an angle selection mechanism to identify high-quality infeasible solutions, thereby enhancing diversity, and the third population adopts an adaptive penalty mechanism to improve feasibility. These populations work cooperatively to identify promising infeasible solutions and navigate infeasible regions to approximate the Pareto front. Finally, five state-of-the-art constrained multi-objective optimization algorithms are compared with NNCP. Out of the total 47 test problems, NNCP outperforms the best-performing baseline algorithm on more than 35 problems, highlighting its superior convergence and diversity capabilities.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"176 ","pages":"Article 113051"},"PeriodicalIF":7.2,"publicationDate":"2025-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143851311","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Efficient and performant Transformer private inference with heterogeneous attention mechanisms","authors":"Peng Hu ,&nbsp;Lei Sun ,&nbsp;Cuiyun Hu ,&nbsp;Xiuqing Mao ,&nbsp;Song Guo ,&nbsp;Jingwen Wang ,&nbsp;Miao Yu","doi":"10.1016/j.asoc.2025.113150","DOIUrl":"10.1016/j.asoc.2025.113150","url":null,"abstract":"<div><div>With the development of large-scale models, Transformer architectures have gained widespread adoption. However, privacy concerns become critical when model inference involves separate ownership of data and model parameters. Existing MPC-based methods for private inference suffer from significant overhead and high latency, where replacing traditional Softmax attention mechanisms with faster alternatives serves as a promising research direction. To achieve a better balance between Transformer model performance and inference speed, we explore the impact of attention mechanisms and attention heads on model performance. First, we found that the performance of the attention mechanism is closely related to the downstream task dataset, and the attention mechanism that is faster on specific datasets can actually achieve better model performance. Additionally, we discovered that for attention mechanisms that experience a performance decline, appropriately restoring the attention heads of the Softmax mechanism can significantly enhance performance. We further observed that the selection of key attention heads under different mechanisms is consistent, providing a basis for designing search strategies adapted to different scenarios. Based on these findings, we propose an MPC-friendly attention mechanism replacement method that enables Transformer private inference to be more efficient and performant. This method incorporates two strategies for selecting and replacing attention mechanisms to address diverse scenario requirements, and the resulting heterogeneous attention mechanism significantly improves the speed of private inference while maximizing model performance. With experiments on different downstream tasks, we demonstrated that our method improves average model performance by 1.94 % compared to standard pre-training models, with an inference speed increase of approximately 3 × . Compared to the state-of-the-art methods, our approach enhances model performance by 1.01–8.03 %, with faster inference speeds. Additionally, when evaluated using comprehensive metrics, our method shows improvements of 4.15 × to 8.97 × compared to other approaches.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"176 ","pages":"Article 113150"},"PeriodicalIF":7.2,"publicationDate":"2025-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143838473","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Adaptive multi-region prediction strategy for dynamic multi-objective optimization","authors":"Tao Zhang ,&nbsp;LinJun Yu ,&nbsp;HuiWen Yu","doi":"10.1016/j.asoc.2025.113072","DOIUrl":"10.1016/j.asoc.2025.113072","url":null,"abstract":"<div><div>This paper proposes a novel adaptive multi-region prediction strategy for dynamic multi-objective optimization problems (DMOPs), which aims to efficiently generate diverse populations in response to environmental changes and facilitate the exploration of the new Pareto front. The strategy consists of two main phases: predictive population initialization and elite-guided resampling. In the predictive population initialization phase, the strategy integrates global exploration and local exploitation. Global exploration divides the population into <span><math><mi>N</mi></math></span> subregions based on population distribution characteristics. For each subregion, the historical information of its center point is used to predict its new position in the next environment, and then a Gaussian mixture model (GMM) is used to sample new individuals based on the position information of all new center points. Local exploitation employs the K-Medoids method to cluster historical Pareto fronts and selects individuals corresponding to the medoids in the decision space as representative individuals. These representative individuals are then used to predict their new locations, followed by Gaussian sampling to generate individuals. The initial predicted population is formed by combining the individuals from global exploration, local exploitation, and randomly generated individuals. In the elite-guided resampling phase, the initial predicted population is evaluated, and top-ranked elite individuals are selected. These elites guide the generation of the final population through Gaussian sampling and Latin Hypercube Sampling (LHS), enhancing solution quality and diversity. The proposed strategy is validated on 14 benchmark problems using MIGD, MHV, R(IGD), and DMIGD metrics. Results demonstrate its better comprehensive performance under varying environmental change intensities (mild, moderate, and severe) compared to existing approaches. Furthermore, its application to a real-world PID controller tuning problem highlights the strategy’s practical potential, showcasing superior performance.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"176 ","pages":"Article 113072"},"PeriodicalIF":7.2,"publicationDate":"2025-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143845252","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Jointly leveraging 1D and 2D convolution on diachronic entity embedding for temporal knowledge graph completion","authors":"Mingsheng He,&nbsp;Lin Zhu,&nbsp;Luyi Bai","doi":"10.1016/j.asoc.2025.113144","DOIUrl":"10.1016/j.asoc.2025.113144","url":null,"abstract":"<div><div>Temporal knowledge graphs (TKGs) model knowledge that dynamically changes over time in the real world, providing effective support for temporal-aware artificial intelligence (AI) applications. However, existing TKGs are far from complete, and their incompleteness significantly affects the performance of downstream applications. Therefore, Temporal Knowledge Graph Completion (TKGC) has become a current research hotspot, which aims to reason potential missing facts based on existing ones. In the widely studied TKGC methods with the implicit representation of temporal information, existing methods that embed temporal information into entity representations can capture the temporal evolution of entities. However, they fail to take the behavioral characteristics of entities across different time units into account, making them challenging to precisely model the fine-grained dynamics of entities. Furthermore, given the powerful expressiveness of Convolutional Neural Networks (CNNs), some TKGC methods have employed the 1D convolution operation to capture global relationships within the embedded quadruple, enabling the learning of explicit knowledge in TKGs and attaining competitive performance for TKGC. Nevertheless, the non-linear and deep features embedded in the entity-relation interaction have not been insufficiently explored. To address these challenges, this paper proposes JointDE, a TKGC model that applies both 1D and 2D convolution operations to the generated diachronic entity embedding, which simultaneously learns the explicit and implicit knowledge in TKGs. The new diachronic entity embedding method explicitly models the inherent attributes of entities and integrates temporal features across different time units, thereby possessing the ability to capture fine-grained entity evolution. More importantly, we construct feature matrices and filters using diachronic entity embeddings and relation embeddings, leveraging an internal 2D convolution mechanism to expand their interactions. This is the first work to learn implicit knowledge embedded in TKGs from a local relationship perspective for TKGC. Experimental results demonstrate that JointDE surpasses several TKGC baseline methods and achieves state-of-the-art performance on three event-based benchmark datasets: ICEWS14, ICEWS05–15, and GDELT. Specifically, JointDE improves Mean Reciprocal Rank (MRR) by 3.17 % and Hits@1 by 5.87 % over the state-of-the-art baseline for entity reasoning.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"176 ","pages":"Article 113144"},"PeriodicalIF":7.2,"publicationDate":"2025-04-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143848447","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}