Knowledge science, engineering and management : 15th international conference, KSEM 2022, Singapore, August 6-8, 2022, proceedings, part ii / / Gerard Memmi, [and four others], (editors)

• Pubblicazione: Cham, Switzerland : , : Springer, , [2022]; ©2022
• Descrizione fisica: 1 online resource (715 pages)
• Disciplina: 658.4038
• Soggetto topico: Knowledge management
• Soggetto non controllato: Mathematics
• Persona (resp. second.): MemmiGerard; YangBaijian; KongLinghe; ZhangTianwei; QiuMeikang
• Note generali: Includes index.
• Nota di contenuto: Intro -- Preface -- Organizations -- Contents - Part II -- Knowledge Engineering Research and Applications (KERA) -- Multi-view Heterogeneous Network Embedding -- 1 Introduction -- 2 Related Work -- 3 Problem Definition -- 4 Methodology -- 4.1 Semantics-Based View Generation -- 4.2 View Preservation and Enhanced View Collaboration -- 4.3 Embedding Fusion -- 4.4 Optimization Objective -- 5 Experiments -- 5.1 Experimental Setup -- 5.2 Link Prediction -- 5.3 Node Classification -- 5.4 Parameter Sensitivity Analysis -- 6 Conclusion -- References -- A Multi-level Attention-Based LSTM Network for Ultra-short-term Solar Power Forecast Using Meteorological Knowledge -- 1 Introduction -- 2 Related Work -- 3 Architecture -- 3.1 Encoder -- 3.2 Decoder -- 4 Experiment: Case Study -- 4.1 Datasets Setting -- 4.2 Experimental Setting -- 4.3 Results and Analysis -- 5 Conclusions -- References -- Unsupervised Person Re-ID via Loose-Tight Alternate Clustering -- 1 Introduction -- 2 Related Work -- 2.1 Clustering-Guided Unsupervised Person re-ID -- 2.2 Camera-Aware Unsupervised Person re-ID -- 3 Methodology -- 3.1 Problem Definition -- 3.2 Loose and Tight Clustering Bounds -- 3.3 Loose-Tight Alternate Clustering -- 3.4 Quality Measurement Based Learning -- 4 Experiments -- 4.1 Datasets and Evaluation Protocol -- 4.2 Implementation Details -- 4.3 Ablation Studies -- 4.4 Comparison with State-of-the-Art Methods -- 4.5 Robustness Evaluation -- 5 Conclusion -- References -- Sparse Dense Transformer Network for Video Action Recognition -- 1 Introduction -- 2 Related Work -- 2.1 CNNs in Action Recognition -- 2.2 Transformer in Action Recognition -- 3 Sparse Dense Transformer Network -- 3.1 Frame Alignment -- 3.2 Patch Crop -- 4 Experiments -- 5 Ablation Experiments -- 6 Conclusion -- References -- Deep User Multi-interest Network for Click-Through Rate Prediction. ; 1 Introduction -- 2 Related Works -- 3 The Proposed Model -- 3.1 Preliminaries -- 3.2 Embedding -- 3.3 Self-Interest Extractor Network -- 3.4 User-User Interest Extractor Network -- 3.5 Prediction and Optimization Objective -- 4 Experiments -- 4.1 Datasets -- 4.2 Competitors and Parameter Settings -- 4.3 Experimental Results -- 4.4 Ablation Study -- 4.5 Parameter Analysis -- 5 Conclusions -- References -- Open Relation Extraction via Query-Based Span Prediction -- 1 Introduction -- 2 Approach -- 2.1 Task Description -- 2.2 Query Template Creation -- 2.3 Encoder -- 2.4 Span Extraction Module -- 2.5 Training and Inference -- 3 Experimental Setup -- 3.1 Datasets -- 3.2 Implementations -- 3.3 Baselines -- 4 Experimental Results -- 4.1 H1: QORE for Multilingual Open Relation Extraction -- 4.2 H2: Zero-shot Domain Transferability of QORE -- 4.3 H3: Few-Shot Learning Ability of QORE -- 5 Conclusion -- References -- Relational Triple Extraction with Relation-Attentive Contextual Semantic Representations -- 1 Introduction -- 1.1 Challenge of Relation Extraction -- 1.2 Our Contribution -- 2 Related Work -- 3 Methodology -- 3.1 Representations of Token and Relation -- 3.2 Relation Prediction -- 3.3 Subject and Object Extraction -- 3.4 Training and Inference -- 4 Experiment and Analysis -- 4.1 Datasets and Settings -- 4.2 Baselines and Evaluation Metrics -- 4.3 Relation Extraction Results -- 4.4 Ablation Study -- 5 Conclusion and Future Works -- References -- Mario Fast Learner: Fast and Efficient Solutions for Super Mario Bros -- 1 Introduction -- 2 Background -- 2.1 Reinforcement Learning -- 2.2 Super Mario Bros Games -- 2.3 Leading Reinforcement Learning Methods -- 2.4 Problems of Previous Methods -- 3 Proposed Methods -- 3.1 Use Accuracy Metrics -- 3.2 Accelerated Training Solution -- 3.3 Target Function Update -- 4 Experiments. ; 4.1 Baseline with Accuracy Check -- 4.2 New Method -- 5 Conclusion -- References -- Few-Shot Learning with Self-supervised Classifier for Complex Knowledge Base Question Answering -- 1 Introduction -- 2 MACL -- 2.1 Overview of the Framework -- 2.2 Algorithm -- 2.3 Objective Function with Reinforcement Learning -- 3 Evaluation -- 3.1 CQA Dataset -- 3.2 Comparison Methods -- 3.3 Implementation Details -- 3.4 Performance Evaluation -- 4 Related Work -- 5 Conclusion -- References -- Data-Driven Approach for Investigation of Irradiation Hardening Behavior of RAFM Steel -- 1 Introduction -- 2 Data Set Construction -- 3 Model Construction and Application -- 3.1 Machine Learning Method -- 3.2 Feature Descriptor -- 3.3 Model Construction -- 3.4 Knowledge Reasoning and Prediction -- 4 Conclusion -- References -- Deep-to-Bottom Weights Decay: A Systemic Knowledge Review Learning Technique for Transformer Layers in Knowledge Distillation -- 1 Introduction -- 2 Related Work -- 3 Methodology -- 3.1 Word Embedding Distillation -- 3.2 Transformer Layer Distillation with Review Mechanism -- 3.3 Prediction Distillation -- 3.4 Total Loss -- 4 Experimental Setup -- 4.1 Experimental Data -- 4.2 Implementation Details -- 4.3 Baseline Methods -- 5 Experimental Results -- 5.1 Main Results -- 5.2 Strategy Comparison -- 6 Conclusions -- References -- Topic and Reference Guided Keyphrase Generation from Social Media -- 1 Introduction -- 2 Related Work -- 3 Methodology -- 3.1 Retriever -- 3.2 Encoder with Heterogeneous Graph -- 3.3 Contextual Neural Topic Model -- 3.4 Topic-Reference-Aware Decoder -- 3.5 Jointly Training -- 4 Experiment Settings -- 4.1 Datasets -- 4.2 Comparisons and Evaluation -- 4.3 Implementation Details -- 5 Results and Analysis -- 5.1 Performance of Keyphrase Generation -- 5.2 Prediction of Present and Absent Keyphrase -- 5.3 Ablation Study. ; 5.4 Influence of the Number of Topics -- 5.5 Case Study -- 6 Conclusion -- References -- DISEL: A Language for Specifying DIS-Based Ontologies -- 1 Introduction -- 2 Background -- 2.1 Domain Information System -- 2.2 Illustrative Example -- 3 Literature Review on Languages for Ontologies -- 3.1 Functional Languages -- 3.2 XML-Based Languages -- 3.3 Other Ontology Languages -- 3.4 Summary -- 4 DISEL Syntax and Support Tool -- 4.1 DISEL Editor Interface Overview -- 4.2 Name and Include Constructs -- 4.3 AtomDomain Construct -- 4.4 Concept -- 4.5 Graph -- 5 Design Decisions -- 6 Discussion -- 7 Conclusion and Future Work -- References -- MSSA-FL: High-Performance Multi-stage Semi-asynchronous Federated Learning with Non-IID Data -- 1 Introduction -- 2 Related Work -- 3 MSSA-FL: Multi-stage Semi-asynchronous Federated Learning -- 3.1 Framework Overview -- 3.2 Combination Module and Multi-stage Training -- 3.3 Semi-asynchronous Training -- 3.4 Model Assignment -- 3.5 Model Aggregation -- 4 Experiment -- 4.1 Experimental Setup -- 4.2 Experimental Results -- 5 Conclusion -- References -- A GAT-Based Chinese Text Classification Model: Using of Redical Guidance and Association Between Characters Across Sentences -- 1 Introduction -- 2 Methodology -- 2.1 Problem Definition -- 2.2 Technical Details of Classification Model -- 3 Evaluation -- 3.1 Dataset Description -- 3.2 Baeline -- 3.3 Experimental Results -- 3.4 Ablation Study -- 4 Conclusion -- References -- Incorporating Explanations to Balance the Exploration and Exploitation of Deep Reinforcement Learning -- 1 Introduction -- 2 Preliminaries -- 2.1 Reinforcement Learning -- 2.2 Variational Inference -- 3 Proposed Method -- 3.1 Network Architecture -- 3.2 Explanation of Actions with Activation Maps -- 3.3 Fusion Activation Maps and States -- 3.4 Encoding the Fused State with Variational Inference. ; 4 Experiments and Results -- 4.1 Environment and Experimental Settings -- 4.2 Comparisons with Benchmark Algorithms -- 4.3 Analysis of Explainability -- 5 Conclusion -- References -- CLINER: Clinical Interrogation Named Entity Recognition -- 1 Introduction -- 2 Proposed Method -- 2.1 Model Design -- 3 Experiments -- 3.1 Baselines and Evaluation Metrics -- 3.2 Experimental Settings -- 3.3 Results and Analysis -- 4 Conclusion -- References -- CCDC: A Chinese-Centric Cross Domain Contrastive Learning Framework -- 1 Introduction -- 2 Related Work -- 2.1 Contrastive Learning -- 2.2 Unsupervised SimCSE -- 2.3 Supervised SimCSE -- 2.4 Sentence Contrastive Learning with PLMs -- 3 CCDC Framework -- 3.1 Cross-Domain Sentences as Hard-Negative Samples -- 3.2 Hard NLI Data Builder -- 3.3 Soft NLI Data Builder -- 4 Experiment -- 4.1 Data Preparation -- 4.2 Training Details -- 4.3 CCDC with One-Domain Training and In-Domain/Out-Domain Testing -- 4.4 CCDC with the Hard/Soft NLI Data Builder -- 5 Analysis -- 6 Conclusion -- 7 Appendix -- 7.1 CCDC with Different PLM and Different Pooling Layer -- 7.2 Case Analysis -- References -- A Multi-objective Optimization Method for Joint Feature Selection and Classifier Parameter Tuning -- 1 Introduction -- 2 Problem Formulation -- 3 The Proposed Approach -- 3.1 Traditional MOGWO -- 3.2 IMOGWO -- 4 Experimental Results and Analysis -- 4.1 Datasets and Setups -- 4.2 Feature Selection Results -- 5 Conclusion -- References -- Word Sense Disambiguation Based on Memory Enhancement Mechanism -- 1 Introduction -- 2 Related Word -- 2.1 Knowledge-Based WSD -- 2.2 Supervised WSD -- 3 Methodology -- 3.1 Task Definition -- 3.2 Model Architecture -- 3.3 Context-Encoder and Gloss-Encoder Units -- 3.4 Memory-Enhancement Unit -- 3.5 Prediction Unit -- 4 Experiments -- 4.1 Dataset -- 4.2 Implementation Details. ; 4.3 Comparison with the State-of-the-Art Baselines.
• Sommario/riassunto: This volume constitutes the refereed proceedings of the 15th International Conference on Knowledge Science, Engineering and Management, KSEM 2022, held in Singapore, during August 6-8, 2022. The 169 full papers presented in these proceedings were carefully reviewed and selected from 498 submissions.
• Titolo autorizzato: Knowledge Science, Engineering and Management
• ISBN: 3-031-10986-4
• Formato: Materiale a stampa
• Livello bibliografico: Monografia
• Lingua di pubblicazione: Inglese
• Record Nr.: 9910585774603321
• Lo trovi qui: Univ. Federico II
• Serie: Lecture Notes in Computer Science