Knowledge Science, Engineering and Management: 16th International Conference, KSEM 2023, Guangzhou, China, August 16–18, 2023, Proceedings, Part III (Lecture Notes in Artificial Intelligence)

This volume set constitutes the refereed proceedings of the 16th International Conference on Knowledge Science, Engineering and Management, KSEM 2023, which was held in Guangzhou, China, during August 16–18, 2023. The 114 full papers and 30 short papers included in this book were carefully reviewed and selected from 395 submissions. They were organized in topical sections as follows: knowledge science with learning and AI; knowledge engineering research and applications; knowledge management systems; and emerging technologies for knowledge science, engineering and management. Preface Organization Keynotes Abstracts Credibility of Machine Learning Through Information Granularity A New Paradigm to Leverage Formalized Knowledge and Machine Learning Recent Advances in Assessing Time Series Similarity Through Dynamic Time Warping ChatGLM: Run Your Own “ChatGPT” on a Laptop Contents – Part III Knowledge Management Systems Explainable Multi-type Item Recommendation System Based on Knowledge Graph 1 Introduction 2 Related Work 3 Methodology 3.1 Graph Transformer Layer 3.2 Graph Convolutional Layer 3.3 Explainable Recommender Layer 4 Experimental Evaluation 4.1 Experimental Setting 4.2 Result Analysis 5 Conclusion References A 2D Entity Pair Tagging Scheme for Relation Triplet Extraction 1 Introduction 2 Related Work 3 Methodology 3.1 Task Definition 3.2 2D Entity Pair Tagging Scheme 3.3 Token-Pair Classifier 4 Experiments 4.1 Datasets and Evaluation Metrics 4.2 Implementation Details 4.3 Baselines 4.4 Overall Results 4.5 Detailed Results on Complex Scenarios 4.6 Results for Different Subtasks 5 Conclusion References MVARN: Multi-view Attention Relation Network for Figure Question Answering 1 Introduction 2 Proposed Method 2.1 Image Representation Learning 2.2 Question Representation Learning 2.3 Multi-view Relation Module 3 Experiments 3.1 Dataset Description 3.2 Experimental Setting 3.3 Experimental Results and Analysis 4 Conclusion References MAGNN-GC: Multi-head Attentive Graph Neural Networks with Global Context for Session-Based Recommendation 1 Introduction 2 Related Work 2.1 RNN-Based or MLP-Based Methods 2.2 GNN-Based Methods 3 Proposed Method 3.1 Problem Statement 3.2 Preliminaries 3.3 Learning Global-Level Item Embedding 3.4 Learning Local-Level Item Embedding 3.5 Multi-information Fusion 3.6 Making Recommendations 4 Experiments 4.1 Experimental Setup 4.2 Comparison with Baselines (RQ1) 4.3 Comparison with Variants of the Proposed Method (RQ2) 4.4 Impact of Multi-head Attention Setting (RQ3) 5 Conclusion References Chinese Relation Extraction with Bi-directional Context-Based Lattice LSTM 1 Introduction 2 Related Work 3 Methodology 3.1 Input Representation 3.2 Context-Based Lattice Encoder 3.3 Cross-Attention Semantic Interaction-Enhanced Classifier 4 Experiments 4.1 Experimental Settings 4.2 Evaluation Results 4.3 Ablation Study 5 Conclusion References MA-TGNN: Multiple Aggregators Graph-Based Model for Text Classification 1 Introduction 2 Related Work 2.1 Machine Learning and Convolutional Neural Network 2.2 Graph Neural Network 2.3 Neighbourhood Aggregation 3 Model Structure 3.1 Node Construction 3.2 Node Updates 3.3 Node Aggregates 4 Experiment 4.1 Dataset 4.2 Baseline 4.3 Experimental Setting 4.4 Dynamic Parameter Setting 4.5 Results 4.6 Ablation Study 5 Conclusion and Future Works References Multi-display Graph Attention Network for Text Classification 1 Introduction 2 Related Work 3 Methods 3.1 Application of Graph Attention Network on Text Classification 3.2 Construction of Multi-display Graph 3.3 Graph-Level Representation Learning 3.4 Inductive Text Classification 3.5 Training 4 Experiments 4.1 Datasets 4.2 Baselines 4.3 Implementation Details 5 Results and Analysis 5.1 Experimental Results 5.2 Ablation Study 5.3 Analysis of Each Layer of the Module 5.4 Visualization 6 Conclusion References Debiased Contrastive Loss for Collaborative Filtering 1 Introduction 2 Related Work 3 Methodology 3.1 Problem Setup 3.2 An Overview of the Proposed Method 3.3 Debiased Contrastive Loss 3.4 Theoretical Analysis 3.5 Model Training 4 Experiments 4.1 Experimental Settings 4.2 Overall Performance Comparison 4.3 Effectiveness Analysis 4.4 Parameter Sensitivity Analysis 5 Conclusion References ParaSum: Contrastive Paraphrasing for Low-Resource Extractive Text Summarization*-4pt 1 Introduction 2 Related Work 3 Method 3.1 PLM-Based Textual Paraphrasing 3.2 Extractive Summarization as Textual Paraphrasing 3.3 Training Paradigm 4 Experiments 4.1 Experimental Settings 4.2 Experimental Results 5 Conclusion References Degree-Aware Embedding and Interactive Feature Fusion-Based Graph Convolution Collaborative Filtering 1 Introduction 2 Related Work 3 Methodology 3.1 Degree-Aware Embedding 3.2 Interaction Feature Fusion 3.3 Feature Representation 3.4 Model Training 4 Experimental Evaluation 4.1 Datasets and Evaluation Metrics 4.2 Baseline Algorithms 4.3 Parameter Settings 4.4 Performance Comparison 4.5 Ablation Analysis 5 Conclusion and Future Work References Hypergraph Enhanced Contrastive Learning for News Recommendation*-4pt 1 Introduction 2 Related Work 3 Methodology 3.1 Preliminary 3.2 News Representation 3.3 Intent Interaction Learning Module 3.4 Hypergraph Structure Learning Module 3.5 Optimization 4 Experiments 4.1 Experimental Settings 4.2 Performance Evaluation 4.3 Ablation Study 4.4 Parameter Analysis 5 Conclusion References Reinforcement Learning-Based Recommendation with User Reviews on Knowledge Graphs 1 Introduction 2 Related Work 2.1 Reinforcement Learning for Recommendation 2.2 User Reviews for Recommendation 3 Methodology 3.1 Rating Prediction Model 3.2 Explorer 3.3 Knowledge Graph Reasoning 3.4 Comparative Model 4 Experiments 4.1 Experimental Setup 4.2 Overall Performance 4.3 Influence of Different Rating Prediction Scores 4.4 Influence of Normalized Quantile 5 Conclusions References A Session Recommendation Model Based on Heterogeneous Graph Neural Network 1 Introduction 2 Related Work 2.1 Traditional Recommender Systems 2.2 Recommendation Based on Graph Neural Networks 2.3 Recommendation System with Sessions 3 The Proposed Method 3.1 Build the Heterogeneous Graph 3.2 Session Embedding 3.3 Rating Prediction and Optimization 4 Experiments and Results 4.1 Datasets 4.2 Comparison Models 4.3 Evaluation Metrics 4.4 Experimental Results and Analysis 5 Conclusion References Dialogue State Tracking with a Dialogue-Aware Slot-Level Schema Graph Approach 1 Introduction 2 Related Work 3 Methodology 3.1 Context and Schema Encoder 3.2 A Two Layers Network: Slot-Token Attention Layer 3.3 A Two Layers Network: Schema Graph Layer 3.4 Value Prediction 4 Experiment 4.1 Data and Experimental Setup 4.2 Result 5 Conclusion References FedDroidADP: An Adaptive Privacy-Preserving Framework for Federated-Learning-Based Android Malware Classification System 1 Introduction 2 Related Work 2.1 FL-Based Malware Classification 2.2 Privacy Protection Methods in Federated Learning 3 Preliminaries 3.1 FL-Based System Model 3.2 Threat Model 3.3 Local Differential Privacy 3.4 Statement of Problems 4 Proposed Model 4.1 Overview 4.2 Privacy Risk Estimation of Android Users’ Sensitive Information 4.3 Privacy Risk-Based Adaptive Differential Privacy Protector ADP 4.4 Privacy Analysis 5 Experiments 5.1 Data and Data Settings 5.2 Model and Training Setup 5.3 Evaluation of the Effectiveness of Privacy Protection 5.4 Evaluation of the Effectiveness of Maintaining the Models’s Utility 5.5 Experimental Discussions 6 Conclusion and Future Work References Multi-level and Multi-interest User Interest Modeling for News Recommendation 1 Introduction 2 Related Works 3 Method 3.1 Problem Formulation 3.2 News Encoder 3.3 Word-Level User Interest Encoder 3.4 News-Level User Interest Encoder 3.5 Higher-Level User Interest Encoder 3.6 Click Predictor 4 Experiments 4.1 Dataset and Experimental Settings 4.2 Main Results 4.3 Ablation Study 4.4 Number of Interest Vectors 5 Conclusion References CoMeta: Enhancing Meta Embeddings with Collaborative Information in Cold-Start Problem of Recommendation 1 Introduction 2 Related Work 3 Method 3.1 Overview 3.2 B-EG: Base Embedding Generator 3.3 S-EG: Shift Embedding Generator 4 Experiments 4.1 Datasets 4.2 Backbones and Baselines 4.3 Experimental Settings 4.4 Experimental Results 5 Conclusion References A Graph Neural Network for Cross-domain Recommendation Based on Transfer and Inter-domain Contrastive Learning 1 Introduction 2 Related Work 3 Method 3.1 Construction of Graphs 3.2 Graph Convolutional Transfer Layer 3.3 Construction of the Contrastive Learning Loss Function 3.4 Rating Prediction and Model Training 4 Experiments and Results 4.1 Datasets 4.2 Baseline Model Comparison 4.3 Model Ablation Experiments 5 Conclusion References A Hypergraph Augmented and Information Supplementary Network for Session-Based Recommendation 1 Introduction 2 Related Works 2.1 Graph Augmentation 2.2 Graph-Based Methods for Session-Based Recommendation 3 Preliminaries 3.1 Notations and Definitions 3.2 Definition of the Global Graph and the Hypergraph 3.3 Hypergraph Convolution Channel 4 The Proposed Approach 4.1 Global Graph Self-supervised Learning Channel 4.2 Hypergraph Augmented Learning Channel 4.3 Readout Function 4.4 Metric Function 4.5 Self-supervised Learning 5 Experiments 5.1 Experimental Setup 5.2 Model Comparison (RQ1) 5.3 Ablation and Effectiveness Analyses (RQ2) 6 Conclusion References Candidate-Aware Attention Enhanced Graph Neural Network for News Recommendation 1 Introduction 2 Problem Formulation 3 Proposed Method 3.1 News Representation 3.2 User Represents Learning 3.3 Click Predictor 4 Experiment 4.1 Datasets and Experimental Settings 4.2 Performance Evaluation 4.3 Ablation Study 4.4 Influence of Hyper-parameters 4.5 Case Study 5 Conclusion References Heavy Weighting for Potential Important Clauses 1 Introduction 2 Preliminaries 3 Implementation and Analysis of SATLC 3.1 Further Improvements 3.2 Implementation Details 4 Experimental Evaluations 4.1 Experiment Preliminaries 4.2 Experimental Results 5 Conclusions References Knowledge-Aware Two-Stream Decoding for Outline-Conditioned Chinese Story Generation 1 Introduction 2 Related Work 3 Methodology 3.1 Problem Formulation 3.2 Overview 3.3 Adding Commonsense Knowledge 3.4 Neural Story Generator 3.5 Exerting the Two-Stream Decoding Mechanism 4 Experiments 4.1 Dataset 4.2 Baselines 4.3 Evaluation Metrics 4.4 Implementation Details 4.5 Results and Analysis 5 Conclusion and Future Work References Multi-path Based Self-adaptive Cross-lingual Summarization 1 Introduction 2 Related Work 3 Our Framework 3.1 Preliminary 3.2 The Proposed Method 4 Experiments 4.1 Datasets 4.2 Experimental Setting 4.3 Experimental Results 5 Conclusion References Temporal Repetition Counting Based on Multi-stride Collaboration 1 Introduction 2 Related Work 2.1 Repetitive Activity Counting 2.2 Video Feature Extraction 2.3 Adaptive Temporal Auto-Correlation 2.4 Temporal Stride Selection 3 Method 3.1 Encoder 3.2 Adaptive Temporal Correlation 3.3 Period Predictor 3.4 Temporal Stride Selection 3.5 Inference 4 Datasets 5 Experiment 5.1 Implementation Details 5.2 Benchmarks and Evaluation Metric 5.3 Evaluation and Comparison 5.4 Ablation Studies 6 Conclusion References Multi-layer Attention Social Recommendation System Based on Deep Reinforcement Learning 1 Introduction 2 Related Work 3 Preliminaries 3.1 Problem Description 4 MAS-DRLRC Model 5 Evaluation 5.1 Parameter Setting and Performance Comparison 5.2 Impact of Embedding Size 6 Conclusion References SPOAHA: Spark Program Optimizer Based on Artificial Hummingbird Algorithm 1 Introduction 2 Related Work 3 Design and Implementation 3.1 Operator Classification 3.2 Rules 3.3 Artificial Hummingbird Model 4 Experiments 4.1 Experimental Results 5 Conclusions and Future Work References TGKT-Based Personalized Learning Path Recommendation with Reinforcement Learning 1 Introduction 2 Related Work 2.1 Mining Personalized Features of Learners 2.2 Knowledge Tracing Based on Deep Learning 2.3 Personalized Learning Path Recommendation 3 Problem Definition 4 Methodology 4.1 The Whole Architecture of Our Model 4.2 Knowledge Tracing Network (KTN) 4.3 Learning Path Recommendation Network (LPRN) 5 Experiment 5.1 Datasets 5.2 Baselines 5.3 Performance 5.4 Ablation Study 6 Conclusion References Fusion High-Order Information with Nonnegative Matrix Factorization Based Community Infomax for Community Detection 1 Introduction 2 Relate Work 2.1 NMF-Based Community Detection Method 2.2 Graph Neural Networks 3 Method 3.1 Graph Attention Networks with High-Order Information 3.2 Graph and Community Level Summary 3.3 Mutual Infomax Learning 4 Experiments 4.1 Datasets 4.2 Comparison Models 4.3 Evaluation Metrics 4.4 Experiments Results 4.5 Ablation Study 5 Conclusion References Multi-task Learning Based Skin Segmentation 1 Introduction 2 Related Work 2.1 Skin Segmentation Algorithm Based on Deep Learning 2.2 Multi-task Learning 2.3 Dynamic Convolution and Transformer 3 Method 3.1 Dynamic Coding 3.2 Distillation Decoding 3.3 Loss Function 4 Experimental Results 4.1 Datasets and Implementation Details 4.2 Result 4.3 Ablation Study 5 Conclusion References User Feedback-Based Counterfactual Data Augmentation for Sequential Recommendation 1 Introduction 2 Related Work 3 UFC4SRec 3.1 Framework Overview 3.2 Sequential Recommendation Rank Score 3.3 Counterfactual Generator 3.4 Recommender-Based Rewards 4 Experiments and Analysis 4.1 Dataset and Metrics 4.2 Parameter Settings 4.3 Results and Analysis 4.4 Hyperparameter Analysis 5 Conclusions References Citation Recommendation Based on Knowledge Graph and Multi-task Learning 1 Introduction 2 Related Work 2.1 Citation Recommendation 2.2 Knowledge Graph 2.3 Multi-task Learning 3 Problem Formulation 4 KMCR Model 4.1 Citation Recommendation 4.2 Paper Knowledge Graph Link Prediction 4.3 Feature Sharing Module 5 Experiments 5.1 Datasets 5.2 Experimental Setup 5.3 Experimental Results 6 Conclusions References A Pairing Enhancement Approach for Aspect Sentiment Triplet Extraction 1 Introduction 2 Related Work 3 Method 3.1 Task Formulation 3.2 Triplet Extraction 3.3 Pair Contrastive Learning 3.4 Training 4 Experiments 4.1 Experimental Setup 4.2 Implementation Details 4.3 Baselines 4.4 Main Results 4.5 Effectiveness of Pairing Information Enhancement 4.6 Comparison of Pairing Strategies 4.7 Pair Feature Visualization 4.8 Case Study 5 Conclusion References The Minimal Negated Model Semantics of Assumable Logic Programs*-4pt 1 Introduction 2 Assumable Logic Program and Its Stable Model Semantics 2.1 Syntax of ALP 2.2 Stable Model Semantics 3 The Minimal Negated Model 3.1 Reverse Reduct 3.2 The Definition of Minimal Negated Models 4 Relation with Stable Models 4.1 The Case of General ALP 4.2 Some Special Cases 5 Conclusion References MT-BICN: Multi-task Balanced Information Cascade Network for Recommendation 1 Introduction 2 Related Work 3 Methodology 3.1 Problem Definition 3.2 The Structure of Multi-task Balanced Information Cascade Network(MT-BICN) 3.3 Training Objective 4 Experiments 4.1 Datasets Description and Baselines 4.2 Experimental Settings 4.3 Performance Comparison 4.4 Ablation Study 5 Conclusion References Author Index