12783nam 22008415 450 991072840010332120230529200123.03-031-33383-710.1007/978-3-031-33383-5(MiAaPQ)EBC30554431(Au-PeEL)EBL30554431(DE-He213)978-3-031-33383-5(BIP)091206136(PPN)270612351(EXLCZ)992680149780004120230529d2023 u| 0engurcnu||||||||txtrdacontentcrdamediacrrdacarrierAdvances in Knowledge Discovery and Data Mining[electronic resource] 27th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2023, Osaka, Japan, May 25–28, 2023, Proceedings, Part IV /edited by Hisashi Kashima, Tsuyoshi Ide, Wen-Chih Peng1st ed. 2023.Cham :Springer Nature Switzerland :Imprint: Springer,2023.1 online resource (360 pages)Lecture Notes in Artificial Intelligence,2945-9141 ;13938Print version: Kashima, Hisashi Advances in Knowledge Discovery and Data Mining Cham : Springer International Publishing AG,c2023 9783031333828 Intro -- General Chairs' Preface -- PC Chairs' Preface -- Organization -- Contents - Part IV -- Scientific Data -- Inline Citation Classification Using Peripheral Context and Time-Evolving Augmentation*-12pt -- 1 Introduction -- 2 Related Work -- 3 Methodology -- 3.1 Cross-Text Attention -- 3.2 Spatial Fusion -- 3.3 Time Evolving Augmentation -- 4 Experiments -- 4.1 Dataset -- 4.2 Implementation Details -- 5 Baselines -- 6 Analysis -- 7 Conclusion -- References -- Social Network Analysis -- Post-it: Augmented Reality Based Group Recommendation with Item Replacement -- 1 Introduction -- 2 Problem Formulation -- 3 STAR3 -- 3.1 Interaction- and Preference-Aware Graph Attention Network -- 3.2 Haptic-Aware Virtual Candidate Item Generator -- 3.3 Social- and Haptic-Aware Recommender -- 3.4 Overall Objective -- 4 Experiments -- 5 Conclusion -- References -- Proactive Rumor Control: When Impression Counts -- 1 Introduction -- 2 Related Work -- 3 Problem Formulation -- 3.1 Influence Model -- 3.2 Influence Block -- 3.3 Problem Definition -- 4 Our Framework -- 4.1 A Baseline -- 4.2 Branch-and-Bound Framework -- 4.3 Computing Upper Bound -- 4.4 Analysis of Solutions -- 5 Progressive Branch-and-Bound -- 6 Experiments -- 6.1 Experimental Settings -- 6.2 Effectiveness Test -- 6.3 Efficiency Test -- 6.4 Scalability Test -- 7 Conclusion -- References -- Spatio-Temporal Data -- Generative-Contrastive-Attentive Spatial-Temporal Network for Traffic Data Imputation -- 1 Introduction -- 2 Related Work -- 3 Preliminaries -- 4 The GCASTN Model -- 4.1 Generative-Contrastive Self-Supervised Learning -- 4.2 Data Augmentation via Two-Fold Cross Random Masking -- 4.3 GCASTN Encoder -- 4.4 GCASTN Decoder -- 5 Experiments -- 5.1 Datasets and Baselines -- 5.2 Experimental Results -- 6 Conclusion -- References.Road Network Representation Learning with Vehicle Trajectories*-12pt -- 1 Introduction -- 2 Problem Definition -- 3 TrajRNE Approach -- 3.1 Spatial Flow Convolution -- 3.2 Structural Road Encoder -- 3.3 TrajRNE Overview -- 4 Experimental Evaluation -- 4.1 Datasets -- 4.2 Baselines -- 4.3 Downstream Tasks and Evaluation Metrics -- 4.4 Experimental Settings -- 4.5 Performance Results -- 4.6 Ablation Study -- 4.7 Parameter Study -- 5 Related Work -- 6 Conclusion -- References -- MetaCitta: Deep Meta-Learning for Spatio-Temporal Prediction Across Cities and Tasks*-12pt -- 1 Introduction -- 2 Problem Statement -- 3 The MetaCitta Approach -- 3.1 Spatial Encoder -- 3.2 Temporal Encoder -- 3.3 Prediction -- 3.4 Training Procedure -- 4 Evaluation Setup -- 4.1 Datasets -- 4.2 Baselines -- 4.3 Experimental Settings -- 5 Evaluation -- 5.1 Comparison with Baselines -- 5.2 Ablation Study -- 5.3 Training Time Comparison -- 6 Related Work -- 7 Conclusion -- References -- Deep Graph Stream SVDD: Anomaly Detection in Cyber-Physical Systems -- 1 Introduction -- 2 Preliminaries -- 2.1 Definitions -- 2.2 Problem Statement -- 3 Methodology -- 3.1 Framework Overview -- 3.2 Embedding Temporal Patterns of the Graph Stream Data -- 3.3 Generating Dynamic Weighted Attributed Graphs -- 3.4 Representation Learning for Weighted Attributed Graph -- 3.5 One-Class Detection with SVDD -- 4 Experiments -- 4.1 Experimental Settings -- 4.2 Experimental Results -- 5 Related Work -- 6 Conclusion -- References -- Texts, Web, Social Media -- Words Can Be Confusing: Stereotype Bias Removal in Text Classification at the Word Level -- 1 Introduction -- 2 Methodology -- 2.1 Problem Formulation -- 2.2 Stereotype Words Detection -- 2.3 Fusion Model Training -- 2.4 Unbiased Prediction -- 3 Experiments -- 3.1 Settings -- 3.2 Classification Performance -- 3.3 Stereotype Word Fairness.3.4 Proportion of Stereotype Words -- 4 Conclusion -- References -- Knowledge-Enhanced Hierarchical Transformers for Emotion-Cause Pair Extraction -- 1 Introduction -- 2 Related Work -- 3 Proposed Method -- 3.1 Overall Architecture -- 3.2 Commonsense Knowledge Injection -- 3.3 Knowledge-Enhanced Clause Encoding -- 3.4 Emotion-Cause Pair Extraction -- 4 Experiments -- 4.1 Datasets and Metrics -- 4.2 Baselines -- 4.3 Implementation Details -- 4.4 Comparison with ECPE Methods -- 5 Conclusion and Future Work -- References -- PICKD: In-Situ Prompt Tuning for Knowledge-Grounded Dialogue Generation -- 1 Introduction -- 2 Related Work -- 3 Methodology -- 3.1 Formal Problem Definition -- 3.2 Contextual Prompting for Knowledge Selection -- 3.3 BART Fine-Tuning for Response Generation -- 4 Experimental Setup -- 4.1 Datasets -- 4.2 Baseline Methods -- 4.3 Evaluation Metrics -- 4.4 Implementation Details -- 5 Empirical Results -- 5.1 Automatic Evaluation -- 5.2 Impact of Prompt Length -- 5.3 Impact of Knowledge Length -- 5.4 Manual Evaluation -- 5.5 Error Analysis -- 6 Conclusion -- References -- Fake News Detection Through Temporally Evolving User Interactions -- 1 Introduction -- 2 Problem Formulation and Data Structure -- 3 Proposed Model -- 3.1 Local Sub-graph Encoding Module -- 3.2 Global Evolution Capturing Module -- 3.3 Neural Hawkes Process Module -- 3.4 Model Training -- 4 Experiment -- 4.1 Datasets -- 4.2 Baseline Methods -- 4.3 Experiment Setting -- 4.4 Performance Comparison -- 4.5 Ablation Study -- 4.6 Early Detection Performance -- 4.7 Case Study -- 5 Related Work -- 6 Conclusion -- References -- Improving Machine Translation and Summarization with the Sinkhorn Divergence*-12pt -- 1 Introduction -- 2 Related Work -- 3 Methodology -- 3.1 Sequence-to-Sequence Model Training -- 3.2 The Proposed Approach: A Contextual Sinkhorn Divergence.4 Experiments -- 4.1 Datasets -- 4.2 Models and Training -- 4.3 Results and Discussion -- 5 Conclusion -- References -- Dual-Detector: An Unsupervised Learning Framework for Chinese Spelling Check -- 1 Introduction -- 2 Method -- 2.1 Overview -- 2.2 Hybrid Mask Strategy -- 2.3 Detector Dec-Err -- 2.4 Candidate Table -- 2.5 Detector Dec-Eva -- 2.6 Training -- 3 Experiments -- 3.1 Datasets and Settings -- 3.2 Main Results -- 3.3 Analysis -- 4 Conclusion -- References -- QA-Matcher: Unsupervised Entity Matching Using a Question Answering Model -- 1 Introduction -- 2 Preliminaries -- 2.1 Question Answering -- 3 Proposed Method -- 3.1 Idea: Solving Entity Matching as Question Answering -- 3.2 Problem Setting -- 3.3 Framework -- 3.4 Retriever -- 3.5 Question and Passage Prompts -- 3.6 QA Classification -- 3.7 Reclassification -- 4 Experiments -- 4.1 Experimental Settings -- 4.2 Results -- 4.3 Ablation Study -- 4.4 Sensitivity Analysis -- 5 Related Work -- 6 Conclusion -- References -- Multi-task Student Teacher Based Unsupervised Domain Adaptation for Address Parsing -- 1 Introduction -- 2 Related Work -- 3 Proposed Methodology -- 3.1 Adaptive Pre-training Using MLM -- 3.2 Student-Teacher Framework -- 3.3 Consistency Regularisation Task -- 3.4 Boundary Detection Task -- 4 Experiments, Data and Results -- 4.1 Data -- 4.2 Experiment Setup -- 4.3 Baselines -- 4.4 Results, Ablation Studies, Parameter Study and Case Study -- 4.5 Training/Inference Time -- 5 Industrial Usecase -- 6 Conclusion and Future Work -- References -- Generative Sentiment Transfer via Adaptive Masking -- 1 Introduction -- 2 Problem Definition -- 3 Methodology -- 3.1 Framework -- 3.2 Adaptive Sentiment Token Masking -- 3.3 Infilling Blanks -- 4 Experiment -- 4.1 Experimental Settings -- 4.2 Quantitative Analysis -- 4.3 Ablation Study -- 4.4 Parameter Sensitivity Analysis -- 5 Conclusion.References -- Unsupervised Text Style Transfer Through Differentiable Back Translation and Rewards -- 1 Introduction -- 2 Related Work -- 3 Methodology -- 3.1 Problem Definition -- 3.2 Shared Encoding -- 3.3 Auto-Encoding -- 3.4 Differentiable Back-Translation -- 3.5 Reinforcement Learning -- 3.6 Learning Technique -- 4 Datasets, Experiments and Results -- 4.1 Datasets -- 4.2 Baselines -- 4.3 Automatic and Human Evaluation -- 5 Analysis -- 5.1 Ablation Studies -- 5.2 Case Study -- 5.3 Error Analysis -- 6 Conclusion and Future Works -- References -- Exploiting Phrase Interrelations in Span-level Neural Approaches for Aspect Sentiment Triplet Extraction*-12pt -- 1 Introduction -- 2 Related Works -- 3 Proposed Method -- 3.1 Contextual Input Representation -- 3.2 Span Construction -- 3.3 Span Filtering -- 3.4 Triplet Construction -- 3.5 Model Training -- 4 Experimental Evaluation -- 4.1 Experimental Setup -- 4.2 Results -- 5 Summary -- References -- What Boosts Fake News Dissemination on Social Media? A Causal Inference View -- 1 Introduction -- 2 Problem Definition -- 3 Our Framework -- 3.1 Preliminary -- 3.2 Causal Feature Representation Learning -- 3.3 Multimodal Covariates Embedding -- 4 Experiment -- 4.1 Evaluation Datasets -- 4.2 Experiment Setting -- 4.3 Main Results -- 4.4 Lexicons Boosting Dissemination -- 5 Related Work -- 6 Conclusion -- References -- Topic-Selective Graph Network for Topic-Focused Summarization -- 1 Introduction -- 2 Related Work -- 2.1 PLM-based Summarization -- 2.2 Topic-Guided Summarization -- 2.3 Graph Neural Network -- 3 Method -- 3.1 Base Topic-Focused Summarization Model -- 3.2 Topic-Arc Recognition -- 3.3 Summarization with Topic-Selective Graph Network -- 3.4 Training -- 4 Experiments -- 4.1 Dataset and Evaluation Metrics -- 4.2 Experimental Setting -- 4.3 Main Results -- 4.4 Ablation Study.4.5 Impact of Topic Node.The 4-volume set LNAI 13935 - 13938 constitutes the proceedings of the 27th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2023, which took place in Osaka, Japan during May 25–28, 2023. The 143 papers presented in these proceedings were carefully reviewed and selected from 813 submissions. They deal with new ideas, original research results, and practical development experiences from all KDD related areas, including data mining, data warehousing, machine learning, artificial intelligence, databases, statistics, knowledge engineering, big data technologies, and foundations.Lecture Notes in Artificial Intelligence,2945-9141 ;13938Artificial intelligenceAlgorithmsEducation—Data processingComputer science—MathematicsComputer visionComputer engineeringComputer networksArtificial IntelligenceDesign and Analysis of AlgorithmsComputers and EducationMathematics of ComputingComputer VisionComputer Engineering and NetworksMathematicsArtificial intelligence.Algorithms.Education—Data processing.Computer science—Mathematics.Computer vision.Computer engineering.Computer networks.Artificial Intelligence.Design and Analysis of Algorithms.Computers and Education.Mathematics of Computing.Computer Vision.Computer Engineering and Networks.006.3Kashima Hisashi908186Ide Tsuyoshi1358929Peng Wen-Chih1358930MiAaPQMiAaPQMiAaPQBOOK9910728400103321Advances in Knowledge Discovery and Data Mining3371743UNINA