Artificial Intelligence Review最新文献

{"title":"On-device large language models: a survey of model compression and system optimization","authors":"Wanyi Chen,&nbsp;Junhao Wang,&nbsp;Yiwei Zhang,&nbsp;Yufan Shi,&nbsp;Tianyi Jiang,&nbsp;Shengxian Zhou,&nbsp;Chenxu Wu,&nbsp;Andi Zhang,&nbsp;Chenyue Zhou,&nbsp;Minxuan Wang,&nbsp;Xinyu Liu,&nbsp;Xiaoshuai Hao,&nbsp;Yinan Wu,&nbsp;Yichen Li,&nbsp;Yuwei Hu,&nbsp;Zhao Cao,&nbsp;Yang Lu,&nbsp;Mengke Li,&nbsp;Yanbiao Ma,&nbsp;Zhiwu Lu,&nbsp;Jungong Han,&nbsp;Yike Guo","doi":"10.1007/s10462-026-11538-1","DOIUrl":"10.1007/s10462-026-11538-1","url":null,"abstract":"<div><p>Large language models are increasingly deployed on device and at the edge, where memory capacity, bandwidth, latency, and privacy requirements dominate system behavior. This survey systematizes the end side stack from algorithms to systems. On the model side, we present a clear taxonomy of quantization, pruning, knowledge distillation, low rank adaptation, and hybrid pipelines, explaining where representative methods belong and how they compose. On the system side, we link these techniques to inference frameworks, compiler and runtime optimizations, kernel fusion, and explicit management of the KV cache. We further propose a unified ALEM protocol, namely Accuracy, Latency, Energy, and Memory, and instantiate it on representative models from 1 to 4 billion parameters to reveal practical trade offs: apply quantization first for memory and time to first token, pair structured pruning with mergeable low rank compensation, and treat the KV cache as a first class subsystem through paging, compression, and eviction. Finally, we outline open problems and directions, including a unified low bit pipeline that couples transform, calibration, and kernel fusion, joint search over structured pruning and distillation, and train and serve unification that collapses sparse, quantized, and low rank parameters into inference ready weights. The goal is a practical bridge from algorithmic compression to resource aligned and reliable on device and edge deployment. https://github.com/LumosJiang/Awesome-On-Device-LLMs: a repository hosting the complete references for Sects. 3–4.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"59 9","pages":""},"PeriodicalIF":18.8,"publicationDate":"2026-05-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-026-11538-1.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148752201","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Explainable artificial intelligence for web application threat detection: a systematic review","authors":"Hamda Fasla,&nbsp;Abid Mehmood","doi":"10.1007/s10462-026-11581-y","DOIUrl":"10.1007/s10462-026-11581-y","url":null,"abstract":"<div><p>Web applications have become integral to critical sectors such as finance, healthcare, and e-commerce, making them prime targets for cyberattacks. Traditional security mechanisms, including Web Application Firewalls (WAFs) and signature-based Intrusion Detection Systems (IDS), struggle to detect zero-day exploits and context-aware attacks, creating the need for more adaptive and intelligent solutions. This survey provides a comprehensive review of Artificial Intelligence (AI)-driven approaches for web application threat detection, with a focus on prevalent attacks such as SQL Injection (SQLi) and Cross-Site Scripting (XSS). The paper systematically analyzes machine learning, deep learning, and transformer-based models, while highlighting the growing importance of Explainable AI (XAI) techniques like SHAP and LIME in improving transparency and analyst trust. We compare state-of-the-art models across widely used datasets (e.g., CSIC 2010, CICIDS2017, UNSW-NB15), evaluate their performance using standard metrics, and identify key limitations such as dataset constraints, high computational overhead, and lack of real-world deployment. Building on this analysis, the survey outlines open research challenges and proposes a future roadmap toward lightweight, scalable, and explainable IDS frameworks. By bridging accuracy with interpretability, this work aims to guide the development of next-generation AI-based systems capable of robustly defending modern web applications against evolving threats.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"59 10","pages":""},"PeriodicalIF":18.8,"publicationDate":"2026-05-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-026-11581-y.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148782884","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Navigating threats in federated recommender systems: a survey of attack models and defense strategies","authors":"Waqar Ali,&nbsp;May Altulyan,&nbsp;Baha Ihnaini,&nbsp;Xiangmin Zhou,&nbsp;Jie Shao","doi":"10.1007/s10462-026-11579-6","DOIUrl":"10.1007/s10462-026-11579-6","url":null,"abstract":"<div><p>Federated Learning (FL) has emerged as a prominent solution for privacy-preserving and efficient learning mechanisms. However, recent investigations have highlighted that the security vulnerabilities of FL remain largely unexplored in Recommender Systems (RS). Research on attack models is significant due to its direct impact on the security of decentralized recommendation algorithms. This survey systematically identifies and analyzes the unique vulnerabilities of Federated Recommender Systems (FedRS) under both classical and Artificial Intelligence (AI)-driven attack models, and presents a structured taxonomy along with synthesized insights that reveal how such threats impact system integrity, user trust, and financial outcomes in recommendation-driven industries. This critical gap is addressed through a comprehensive analysis of attack models and state-of-the-art defense strategies in FedRS. Existing attack possibilities in RS, and specifically in FedRS, are collected and analyzed from multiple perspectives. Additionally, vulnerabilities identified in mainstream FL research are examined for their impact within the RS context. Based on this systematic investigation, a novel taxonomy of threat and defense models is proposed, offering a structured classification of attack vectors and an in-depth examination of corresponding defense strategies. This study covers a wide range of attack types, including recent AI-empowered federated attack models and classical attacks, and provides insights into the ethical implications and associated business risks. In addition to the theoretical investigation, a public repository of relevant codebases, metrics and datasets has been compiled to promote future research and facilitate reproducibility in this rapidly evolving domain.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"59 8","pages":""},"PeriodicalIF":18.8,"publicationDate":"2026-05-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-026-11579-6.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148466146","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Addressing catastrophic forgetting in class-incremental learning—a survey","authors":"John Bako,&nbsp;Jugal Kalita","doi":"10.1007/s10462-026-11575-w","DOIUrl":"10.1007/s10462-026-11575-w","url":null,"abstract":"<div><p>Class-Incremental Learning (CIL) enables models to learn new classes over time without forgetting previously acquired knowledge—a process often hindered by <i>Catastrophic Forgetting (CF)</i>. This paper provides a comprehensive and structured overview of the CIL landscape, beginning with a detailed examination of its core variants—Task-, Domain-, and Class-Incremental Learning—and explaining how CIL fits within the broader structure of incremental learning by clarifying the relationships among its major variants and the shared challenges they address. This perspective establishes a consistent conceptual foundation for understanding how different forms of incremental learning relate to continual learning objectives more broadly. The paper then reviews CF in depth, including formal definitions, underlying mechanisms, and evaluation protocols. A central contribution is a refined taxonomy of CIL methods, encompassing replay-based, regularization-based, parameter-isolation, hybrid, and large language model (LLM)-based approaches. Each category is analyzed in terms of method variations, applications, trade-offs, and emerging trends. Quantitative and qualitative comparisons highlight differences in scalability, interpretability, and adaptability to various domains. We also identify open challenges, including bias toward recent classes, memory and compute constraints, and the need for stronger theoretical cohesion in continual adaptation. Looking forward, we outline grounded future directions such as advances in self-supervised and generative replay mechanisms, attention-based transformers, and neuro-symbolic integrations. These directions reflect promising long-term avenues for rethinking and improving representational stability, efficiency, and generalization in class-incremental learning.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"59 8","pages":""},"PeriodicalIF":18.8,"publicationDate":"2026-05-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-026-11575-w.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148466513","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Paving the way towards kinematic assessment using monocular video: a preclinical benchmark of state-of-the-art deep-learning-based 3d human pose estimators against inertial sensors in daily living activities","authors":"Mario Medrano-Paredes,&nbsp;Carmen Fernández-González,&nbsp;Hichem Saoudi,&nbsp;Francisco-Javier Díaz-Pernas,&nbsp;Javier González-Alonso,&nbsp;Mario Martínez-Zarzuela","doi":"10.1007/s10462-026-11559-w","DOIUrl":"10.1007/s10462-026-11559-w","url":null,"abstract":"<div><p>Advances in machine learning and wearable sensors offer new opportunities for capturing and analyzing human movement outside specialized laboratories. Accurate assessment of human movement under real-world conditions is essential for telemedicine, sports science, and rehabilitation. This preclinical benchmark compares monocular video-based 3D human pose estimation models with inertial measurement units (IMUs), leveraging the VIDIMU dataset containing a total of 13 clinically relevant daily activities which were captured using both commodity video cameras and five IMUs. During this initial study only healthy subjects were recorded, so results cannot be generalized to pathological cohorts. Joint angles derived from state-of-the-art deep learning frameworks (MotionAGFormer, MotionBERT, MMPose 2D-to-3D pose lifting, and NVIDIA BodyTrack) were evaluated against joint angles computed from IMU data using OpenSim inverse kinematics following the Human3.6 M dataset format with 17 keypoints. Among them, MotionAGFormer demonstrated superior performance, achieving the lowest overall RMSE (9.27° ± 4.80°) and MAE (7.86° ± 4.18°), as well as the highest Pearson correlation (0.86 ± 0.15) and the highest coefficient of determination R² (0.67 ± 0.28). The results reveal that both technologies are viable for out-of-the-lab kinematic assessment. However, they also highlight key trade-offs between video- and sensor-based approaches including costs, accessibility, and precision. This study clarifies where off-the-shelf video models already provide clinically promising kinematics in healthy adults and where they lag behind IMU-based estimates while establishing valuable guidelines for researchers and clinicians seeking to develop robust, cost-effective, and user-friendly solutions for telehealth and remote patient monitoring.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"59 8","pages":""},"PeriodicalIF":18.8,"publicationDate":"2026-05-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-026-11559-w.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148466512","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Intelligent driver monitoring systems: a survey of drowsiness detection technologies for road safety","authors":"Osama F. Hassan,&nbsp;Ahmed F. Ibrahim,&nbsp;M. A. Makhlouf,&nbsp;B. Hafiz,&nbsp;Ahmed Gomaa","doi":"10.1007/s10462-026-11505-w","DOIUrl":"10.1007/s10462-026-11505-w","url":null,"abstract":"<div><p>Driver drowsiness is a recognized contributor to road crashes. This review provides a structured synthesis of driver-monitoring methods spanning image/video, physiological signals (EEG, ECG, PPG, EOG), vehicle behavior, and multimodal fusion. We conducted a systematic search of major bibliographic databases with a PRISMA workflow covering 2020–2025, screened full texts against inclusion criteria, and coded datasets, validation protocols, metrics, and deployment factors (details in Methods). Across modalities, vision has shifted from classical pipelines to deep learning (including Vision Transformers), while physiological approaches offer robustness under poor visibility at the cost of additional hardware and comfort constraints; multimodal fusion improves stability but raises compute and integration burden. Reported performance remains highly protocol-dependent: subject-dependent or within-dataset validation can inflate headline numbers, whereas subject-independent and cross-dataset tests are rarer and typically yield lower—but more realistic—estimates. We highlight and address a benchmarking gap: heterogeneity in datasets and protocols makes cross-paper accuracy comparisons misleading; we therefore standardize reporting and emphasize operational metrics. We recommend prevalence-aware evaluation with precision–recall analysis (PR-AUC), operating-point metrics (e.g., recall at fixed false-alarm rates), calibration assessment, and time-to-detect. We also discuss privacy, regulation, and deployment considerations (edge/on-device processing). Limitations of the evidence base include scarce large-scale on-road studies, inconsistent ground-truthing, and limited cross-demographic validation. We propose a field-specific research agenda and a shared-authority framework in which DMS negotiates control between human and autonomy with verifiable safety metrics. Together, these contributions provide a consolidated map of the field and practical guidance for robust, real-world evaluation.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"59 6","pages":""},"PeriodicalIF":13.9,"publicationDate":"2026-05-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-026-11505-w.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147938265","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Quantum adversarial machine learning: from classical adaptations to quantum-native methods","authors":"Roozbeh Razavi-Far,&nbsp;Mohammad Meymani,&nbsp;Erfan Mahmoudinia,&nbsp;Dorsa Vazirzade,&nbsp;Peyman Paknezhad,&nbsp;Fateme Ghasemi,&nbsp;Saeed Saravani,&nbsp;Somayeh Nikkhoo,&nbsp;Kimia Haghjooei","doi":"10.1007/s10462-026-11578-7","DOIUrl":"10.1007/s10462-026-11578-7","url":null,"abstract":"<div><p>Machine learning has revolutionized numerous industrial domains. Despite recent advances, machine learning models remain vulnerable to adversarial threats. Adversarial machine learning is a field that studies these vulnerabilities to build robust machine learning models. Quantum machine learning is an interdisciplinary field that bridges quantum computing and classical machine learning. While quantum machine learning shows potentials to outperform classical machine learning in complex tasks such as regression, classification, and generative modeling, it remains vulnerable to adversarial attacks. Given the recent advancements in quantum computing and machine learning, the quantum adversarial machine learning field has emerged to study the vulnerabilities of quantum machine learning, possible attacks, and novel quantum-enhanced defense strategies. In this survey, we provide a detailed overview on quantum adversarial machine learning and explore the existing attacks and countermeasures. We also review the theoretical underpinnings of this area, emerging trends, and critical challenges.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"59 8","pages":""},"PeriodicalIF":18.8,"publicationDate":"2026-05-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13337728/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148403204","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"LLM agents security duality: a comprehensive survey of self-security and empowered cybersecurity","authors":"Yiwei Xu,&nbsp;Yong Zhuang,&nbsp;Xuanming Liu,&nbsp;Tian Zhang,&nbsp;Bowen Xiao,&nbsp;Xiaoyang Xu,&nbsp;Delong Jiang,&nbsp;Juan Wang,&nbsp;Hongxin Hu","doi":"10.1007/s10462-026-11563-0","DOIUrl":"10.1007/s10462-026-11563-0","url":null,"abstract":"<div><p>Large language model (LLM) agents are rapidly being integrated into real-world systems. Their autonomy and tool-use capabilities generate substantial value while simultaneously expanding the security attack surface. This survey provides a comprehensive overview of the opportunities and challenges of LLM agents in security, focusing on two core areas: (1) threats to LLM agents themselves and corresponding mitigation strategies (LLM agents self-security), and (2) the role of LLM agents in empowering the cybersecurity lifecycle across offense and defense (LLM agents empowered cybersecurity). We first examine the internal and external attack surfaces of agents, propose a taxonomy organized by threat sources, and analyze associated mitigations and evaluation frameworks. We then investigate how agent capabilities are applied in cybersecurity practice and present, to our knowledge, the first agent-empowerment framework aligned with the full cyber offense–defense lifecycle. By systematically surveying these two areas, we are the first to highlight a positive feedback synergy between LLM agents self-security and empowered cybersecurity, offering new insights for the advancement of both. We further identify current limitations and outline promising directions for future research. The insights provided aim to catalyze the coordinated development of LLM agents self-security and agent empowered cybersecurity, paving the way for more capable and robust agent applications.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"59 8","pages":""},"PeriodicalIF":18.8,"publicationDate":"2026-05-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-026-11563-0.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148380947","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"The application of federated learning in breast cancer diagnosis and treatment: a survey","authors":"Huafeng Wu,&nbsp;Mohsen Ghorbian","doi":"10.1007/s10462-026-11574-x","DOIUrl":"10.1007/s10462-026-11574-x","url":null,"abstract":"<div><p>Breast cancer (BC) is one of the most common malignancies among women and is associated with complex mechanisms in cell signalling pathways and tumour heterogeneity. These complexities have made the early screening and selection of effective treatments a significant challenge. Federated Learning (FL), a new approach in artificial intelligence, enables the analysis of multicenter data without requiring physical data aggregation. This feature leverages privacy-preserving mechanisms, such as secure aggregation, to improve model accuracy while preserving patient data confidentiality. In the field of BC, FL can significantly improve the accuracy of imaging-based diagnostic algorithms and the efficiency of personalized treatment methods by integrating data from multiple medical institutions. Despite these advantages, the application of FL in the field of BC is associated with specific challenges, including data heterogeneity across institutions, high computational and communication overhead caused by repeated local model training and iterative federated communication rounds, and difficulty in ensuring model generalizability in diverse clinical settings, which pose critical practical limitations for widespread clinical adoption. This review article systematically examines the applications of FL in BC. In the analysis of the articles, it was found that the most significant focus is on improving the accuracy and precision of results, Precision and Accuracy Enhancement (PAE) with 29%, followed by Data Confidentiality and Protection (DCP) with 24%, and finally Model Robustness and Generalization (MRG) with 19%. The findings of this study indicate that FL can improve screening accuracy, recommend more targeted treatments, and, at the same time, prevent the disclosure of sensitive patient information. Ultimately, this research can help oncology and data science researchers and experts better understand the potential and challenges of FL and develop new strategies to improve the quality of clinical care in BC.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"59 8","pages":""},"PeriodicalIF":18.8,"publicationDate":"2026-05-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-026-11574-x.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148381034","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Intention mining: a systematic literature review and research agenda","authors":"Ramona Elali,&nbsp;Rébecca Deneckère,&nbsp;Elena Kornyshova","doi":"10.1007/s10462-026-11549-y","DOIUrl":"10.1007/s10462-026-11549-y","url":null,"abstract":"<div><p>Intention Mining is an Artificial Intelligence field used to discern users’ intentions. It is widely used nowadays across various fields like Robotics, Network Forensics, Security, Bioinformatics, Learning, Map Visualization, Information Retrieval, Games, and much more. Despite the promising research in this area, it doesn’t exist any detailed overview of the existing research contributions. In this paper, we present a systematic literature review using a comparison framework that aims to highlight the different dimensions in the domain of Intention Mining. We study 137 existing contributions in Intention Mining, their underlying purposes, and the various techniques and tools employed. We used Kitchenham’s research method, including the planning, conducting, and reporting of the review phases. Additionally, we outline a research agenda highlighting several future directions, for instance, the need to formalize the intention concept, consider heterogeneous data, or provide runtime recommendations. This work also provides tips for users of artificial intelligence across various application domains.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"59 8","pages":""},"PeriodicalIF":18.8,"publicationDate":"2026-05-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-026-11549-y.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148381033","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}