Towards Open and Trustworthy Digital Societies : 23rd International Conference on Asia-Pacific Digital Libraries, ICADL 2021, Virtual Event, December 1–3, 2021, Proceedings

This book constitutes the refereed proceedings of the 23rd International Conference on Asia-Pacific Digital Libraries, ICADL 2021, which was held in December 2021. Due to COVID-19 pandemic the conference was held virtually. The 17 full, 14 short, and 5 practice papers presented in this volume were carefully reviewed and selected from 87 submissions. The papers were organized in topical sections named: Knowledge Discovery from Digital Collections; Search for Better User Experience; Information Extraction; Multimedia; Text Classification and Matching; Data Infrastructure for Digital Libraries; Data Modeling; Neural-based Learning. Preface Organization Contents Knowledge Discovery from Digital Collections Diachronic Linguistic Periodization of Temporal Document Collections for Discovering Evolutionary Word Semantics 1 Introduction 2 Problem Definition 3 Temporal Word Embeddings 3.1 Learning Static Embeddings 3.2 Learning Dynamic Embeddings 4 Document Collection Periodization 4.1 Scoring 4.2 Periodization 4.3 Analysis of Time Complexity 5 Periodization Effectiveness 5.1 Datasets 5.2 Experimental Settings 5.3 Analyzed Methods 5.4 Test Sets 5.5 Evaluation Metrics 5.6 Evaluation Results 6 Related Work 6.1 Text Segmentation 6.2 Temporal Word Embeddings 7 Conclusion References Joint Model Using Character and Word Embeddings for Detecting Internet Slang Words 1 Introduction 2 Related Works 2.1 Word and Character Embedding 2.2 ELMo 2.3 BERT 2.4 Detection of Internet Slang Words 3 Internet Slang Corpus 3.1 New Semantic Words 3.2 New Blend Words 3.3 Word-Formation Subcategories 3.4 Corpus Construction 4 Method 4.1 Joint Model Using Character and Word Embeddings 4.2 ELMo Representation 4.3 Subword-Level Representation Based on BERT 5 Experiments 5.1 Dataset and Settings 5.2 Baseline Methods 5.3 Experimental Results 5.4 Discussion 6 Conclusion References Knowledge Discovery from the Digital Library’s Contents: Bangladesh Perspective 1 Introduction 2 Literature Review 2.1 Concept of Knowledge Discovery 2.2 Tools of Knowledge Discovery 2.3 The Current State of Digital Library Practice in Bangladesh 3 Objectives of the Study 4 Research Questions (RQs) 5 Research Methodology 6 Findings and Discussion 6.1 Demographic Profile of the Respondents 6.2 Designation of the Respondents 6.3 Competencies of Respondents in Digital Library Management 6.4 The Tools and Technologies Needed to Discover Knowledge from the DL Contents 6.5 The Difficulties and Obstacles of Discovering Knowledge 6.6 Ways of Overcoming the Barriers and Challenges of KD from Contents of DL 6.7 The Most Commonly Used Discovery Tool/software in the Libraries of Bangladesh 7 The Technique of Discovering Knowledge as Per the Library Professionals in Bangladesh 8 Limitations and Direction for Future Research 9 Conclusion References DataQuest: An Approach to Automatically Extract Dataset Mentions from Scientific Papers 1 Introduction 2 Related Work 3 Methodology 3.1 Dataset-Sentence Classification 3.2 Dataset Mention Extraction 4 Results and Analysis 4.1 Analysis 4.2 Error Analysis 5 Conclusion and Future Work References Embedding Transcription and Transliteration Layers in the Digital Library of Polish and Poland-Related News Pamphlets 1 Introduction 2 Related Work and Background Information 3 Embedding Transcription and Transliteration Layers in a PDF File 4 Conclusions and Future Plans References Search for Better User Experience A Qualitative Evaluation of User Preference for Link-Based vs. Text-Based Recommendations of Wikipedia Articles 1 Introduction 2 Background 2.1 Link-Based Similarity Measure 2.2 Text-Based Similarity Measure 2.3 Related Work 3 Methodology 3.1 Study Design 4 Results 4.1 Primary Findings 4.2 Secondary Findings 5 Discussion 6 Conclusion References Narrative Query Graphs for Entity-Interaction-Aware Document Retrieval 1 Introduction 2 Related Work 3 Narrative Query Graphs 4 Narrative Document Retrieval 4.1 Document Graph Extraction 4.2 Document Retrieval 4.3 Prototype Design 4.4 Retrieval with Variable Nodes 5 System Evaluation and User Study 5.1 Retrieval Evaluation 5.2 User Interviews 5.3 Questionnaire 5.4 Performance Analysis 6 Discussion and Conclusion References BookReach-UI: A Book-Curation Interface for School Librarians to Support Inquiry Learning 1 Introduction 2 Related Work 2.1 Teaching-Material Curation by School Librarians 2.2 Support Tools for School Librarians 3 Proposed User Interface: BookReach-UI 3.1 Overall Design 3.2 Detailed Features 4 Experiment 5 Conclusion References SmartReviews: Towards Human- and Machine-Actionable Representation of Review Articles 1 Introduction 2 Related Work 3 Approach 4 Implementation 4.1 Section Types 4.2 Implementation per SmartReview Dimension 5 Evaluation 6 Discussion 7 Conclusions References A Large-Scale Collection Reselection Method Oriented by Library Space Rebuilding Based on the Statistics of Zero-Borrowing Books 1 Introduction 2 Related Work 3 Condition Establishment and Data Generation 3.1 The Establishment of the Conditions for Selecting Books 3.2 Book Extraction Data Generation 4 Results and Analysis 4.1 Analysis of Readers Borrowing Needs Guarantee 4.2 Analysis of Disciplines Distribution Reasonability 4.3 Analysis of Key Disciplines Guarantee 5 Discussion 5.1 The Book Extraction Data by the Unit of “volume” Can Automatically Generate the Number of Copies Reserved for Each Kind of Book 5.2 The Book Extraction Data Should Be Verified by the Rationality of Collection Disciplines Structure 6 Conclusion References Information Extraction Automatic Cause-Effect Relation Extraction from Dental Textbooks Using BERT 1 Introduction 2 Background and Related Work 3 Methodology 3.1 Dataset and Data Preprocessing 3.2 Cause-Effect Sentence Classification 3.3 Cause-Effect Relation Extraction 4 Experiments, Results, and Discussion 4.1 Cause-Effect Sentence Classification 4.2 Cause-Effect Relation Extraction 5 Future Directions 5.1 Relation Linking 5.2 Application in Dental Quiz Generation 6 Conclusions References Multilingual Epidemic Event Extraction 1 Introduction 2 Related Work 3 Dataset 4 Experiments 4.1 Supervised Learning Experiments 4.2 Semi-supervised Learning Experiments 5 Conclusions References Supervised Learning of Keyphrase Extraction Utilizing Prior Summarization 1 Introduction 2 Related Work 3 Methodology 3.1 Motivations 3.2 Candidate Phrase Selection 3.3 BERT Keyphrase-Rank (BK-Rank) 4 Experiments 4.1 Datasets 4.2 Baseline Methods 4.3 Experimental Details 4.4 Performance Comparison 5 Conclusion References Bert-Based Chinese Medical Keyphrase Extraction Model Enhanced with External Features 1 Introduction 2 Related Work 3 Methodology 3.1 Data Preprocessing 3.2 Model Architecture 4 Experiments and Results 4.1 Experimental Design 4.2 Results 5 Conclusions References Multimedia A Novel Approach to Analyze Fashion Digital Library from Humanities 1 Introduction 2 Related Works 2.1 Fashion Trend Analysis 2.2 Digital Fashion Archives 3 CAT STREET: Chronicle Archive of Tokyo Street-Fashion 4 Retrieving Everyday Fashion Trends 4.1 Fashion Style Clustering Model 4.2 Verification 5 Analyzing Fashion Trend Patterns 5.1 Simultaneous Emergence of Fashion Styles 5.2 Emergence of Fashion Styles at Different Types in the Streets 5.3 Discussion 6 Conclusions References The Effect of Library Virtual Tour on Library Image Construction: Study on Perpustakaan BPK RI 1 Introduction 2 Theoretical Background 2.1 Virtual Tour 2.2 Perpustakaan BPK RI Virtual Tour 2.3 Library Image 3 Research Methodology 3.1 Research Models and Hypotheses 3.2 Data Collection 3.3 Data Measurement 4 Data Analysis 4.1 Descriptive Analysis of Research Variables 4.2 Pearson Correlation Test 4.3 Simple Linear Regression Test 5 Advanced Discussion and Analysis 5.1 Virtual Tour and Media Social Are Effective Media to Attract Young Generation to Visit and Use Information Institutions (Libraries, Archives, Museums and Galleries) 5.2 Perpustakaan BPK RI Virtual Tour has a Very Good Quality and the Image of the Perpustakaan BPK RI has a Very Good Image Based on User Rating 5.3 There is a Strong Relationship with Positive Direction Between Library Virtual Tour with Library Image 5.4 Library Virtual Tour Affects Library Image 5.5 Constraints/Barriers in Utilizing the Perpustakaan BPK RI Virtual Tour 6 Conclusions and Suggestions References Exploring the Research Utility of Fan-Created Data in the Japanese Visual Media Domain 1 Introduction 2 Data Structures and Vocabularies of Community Data 2.1 The Visual Novel Database 2.2 The Anime Characters Database 2.3 AnimeClick 3 Discussion and Impact of the Data Analysis 3.1 Impact on the JVMG Ontology 3.2 Impact on Researchers 4 Conclusion and Future Work References Interactive Curation of Semantic Representations in Digital Libraries 1 Introduction 2 Related Work 3 Computer-Assisted Curation of Document Landscapes 4 Evaluation 5 Conclusion References Text Classification and Matching Automating the Choice Between Single or Dual Annotation for Classifier Training 1 Introduction 2 Related Work 2.1 Efficient Construction of Training Data 2.2 Multi-armed Bandit Problem 3 Selection Algorithm 3.1 Overview 3.2 Algorithm 4 Experiment 4.1 Experimental Settings 4.2 Experimental Results 4.3 Discussion 5 Conclusion References Integrating Semantic-Space Finetuning and Self-Training for Semi-Supervised Multi-label Text Classification 1 Introduction 2 Related Work 3 Preliminaries 3.1 Problem Formulation 3.2 Self-training 4 Proposed Approach 4.1 Semantic Label Propagation Algorithm (SemLPA) 4.2 Classifier Finetuning 4.3 High-Confidence Prediction Selection 5 Experiments 5.1 Datasets 5.2 Baselines 5.3 Implementation Details 5.4 Results 5.5 Exploring the Impact of the Semantic-Space on SemLPA 5.6 Comparing the High-Confidence Set Derived from Different Models 6 Conclusion References Named Entity Recognition Architecture Combining Contextual and Global Features 1 Introduction 2 Related Work 2.1 Named Entity Recognition 2.2 Embeddings 3 Methodology 3.1 GCN as Global Embeddings 3.2 XLNet as Contextual Embeddings 3.3 Joint Architecture 4 Experimental Setup 4.1 Dataset 4.2 Implementation Details 4.3 Metrics 5 Results 6 Conclusion and Future Work References A Consistency Analysis of Different NLP Approaches for Reviewer-Manuscript Matchmaking 1 Introduction 2 Problem Formulation 3 Conference Dataset Description 4 Methodologies Implemented and Result Analysis 4.1 Experimental Setup 4.2 Evaluation Method 4.3 Methodologies Description and Implementation 4.4 Result Analysis 5 Conclusion and Future Work References Data Infrastructure for Digital Libraries When Expertise Gone Missing: Uncovering the Loss of Prolific Contributors in Wikipedia 1 Introduction 2 Related Work 3 Dataset Description 4 Feature Space Design 4.1 Activity Features 4.2 Quality Features 4.3 Linguistic Features 4.4 Feature Correlation 5 Classification of Editors 6 Discussion 7 Conclusion References Federating Scholarly Infrastructures with GraphQL 1 Introduction 2 Related Work 3 Approach 3.1 Federated Access to Scholarly Infrastructures 3.2 Methods for Linking Scholarly Knowledge 4 Discussion 5 Conclusion References Simple DL: A Toolkit to Create Simple Digital Libraries 1 Introduction 2 Related Work 3 System Design 3.1 Features 3.2 Storage Layout 3.3 Import, Index and Generate 3.4 Web Management Interface 3.5 Users and Entities 3.6 Comments and Submissions 3.7 Information Retrieval System 4 Case Studies 4.1 Emandulo 4.2 Digital Bleek and Lloyd 4.3 NDLTD Document Archive 5 Reflections References Towards Approximating Population-Level Mental Health in Thailand Using Large-Scale Social Media Data 1 Introduction 2 Methodology 2.1 Mental State Classification 2.2 Aggregation 3 Datasets 3.1 Datasets for Training Classifiers 3.2 Social Media Data 3.3 Ground-Truth Mental Illness Cases 4 Experiments and Results 4.1 Mental State Classification 4.2 Population-Level Mental Health 5 Potential Applications 6 Conclusions References Analysis of the Usefulness of Critique Documents on Musical Performance: Toward a Better Instructional Document Format 1 Introduction 2 Related Work 2.1 Music Database for Research 2.2 Document Format 3 Materials and Methods 3.1 Materials 3.2 Method: Statistical Models for Analysis 4 Results 4.1 Usefulness Evaluation and Types Annotation 4.2 Multilevel Modeling 5 Discussion 6 Conclusion References Data Modeling Modelling Archaeological Buildings Using CIDOC-CRM and Its Extensions: The Case of Fuwairit, Qatar 1 Introduction 2 Related Work 3 Preliminaries 3.1 The Archaeology of Buildings 3.2 The CIDOC CRM and Its Extensions Related to the Archaeology of Buildings 4 The Origins of Doha and Qatar Project, and the Archaeological Works at Fuwairit 5 Describing Archaeological Buildings in Fuwairit and Modelling in CIDOC-CRM and CRMba 5.1 Archaeological Building Components, Construction Techniques and Materials 5.2 Describing the Visual Representation of a Building 5.3 On the Provenance of Information About Archaeological Buildings 6 Conclusions and Future Work References Integrating Heterogeneous Data About Quebec Literature into an IFLA LRM Knowledge Base 1 Introduction 2 Related Work 3 Methodology 3.1 Datasets 3.2 Competency Questions 3.3 The IFLA LRM Model 3.4 Data Challenges 4 IFLA LRM Enrichment 5 Architecture 5.1 Preprocessing 5.2 Translation Based on Competency Questions 5.3 Alignment Strategies 5.4 The Unified Knowledge Base 6 Results 6.1 Descriptive Statistics of the Datasets 6.2 Alignment Evaluation 6.3 Knowledge Base Evaluation 7 Discussion and Conclusion References Evaluating the Robustness of Embedding-Based Topic Models to OCR Noise 1 Introduction 2 Related Work 3 Methodology 3.1 Datasets 3.2 Model Training and Word Embeddings 3.3 Evaluation Measures 4 Results and Discussion 4.1 Performance on Real Noise 4.2 Performance on Synthetic Noise 5 Conclusions References Pattern-Based Acquisition of Scientific Entities from Scholarly Article Titles 1 Introduction 2 Related Work 3 Preliminary Definitions 4 Tool Description 4.1 Formalism 4.2 Rule-Based Processing Workflow 5 Evaluation 5.1 Quantitative Analysis: Scientific Concept Extraction Precision 5.2 Qualitative Analysis: Top N Terms 6 Conclusion and Future Directions References Enriching the Metadata of a Digital Collection to Enhance Accessibility: A Case Study at Practice in Kyushu University Library, Japan 1 Introduction 2 Rare Materials Digital Archive at Kyushu University Library 2.1 Outline of the Digital Collections 2.2 Rare Materials Digital Archive 2.3 Total Number of Accesses of the Rare Materials Digital Archive 3 Metadata Enrichment 3.1 Enriching Metadata 3.2 Number of Accesses 3.3 Discussion 4 Conclusion References Neural-Based Learning PEERAssist: Leveraging on Paper-Review Interactions to Predict Peer Review Decisions 1 Introduction 2 Related Work 3 Data Description 3.1 Analysis and Potential Use Cases 4 Approach Description 4.1 Data Pre-processing 4.2 Embedding Layer 4.3 VADER Sentiment Analyzer 4.4 Deep Learning-Based Features Extraction 4.5 Cross-Attention for Paper-Review Interaction 4.6 Fusion and Accept/Reject Decisions 5 Experimental Setup 5.1 Implementation Details 5.2 Comparing Systems 5.3 Results and Discussion 5.4 Ablation Study 5.5 Observation 5.6 Error Analysis 6 Conclusion and Future Work References ContriSci: A BERT-Based Multitasking Deep Neural Architecture to Identify Contribution Statements from Research Papers 1 Introduction 2 Related Work 3 Task Definition 4 Dataset Description 5 Data Preprocessing 5.1 Combining Incomplete Sentences in the Stanza File 5.2 Extraction of Main Section and Sub Section Titles 5.3 Extracting Previous and Next Sentence 6 Data Analysis After Pre-processing 6.1 Analysis of Contribution Sentences in Sections 6.2 Analysis of Contribution Sentences Having Citation 7 Methodology 7.1 ContriSci Model 7.2 ContriSci Architecture Description 7.3 Data for Scaffold Tasks 7.4 Training Procedure 7.5 Loss 8 Evaluation 8.1 Experimental Setup 8.2 Hyperparameters 8.3 Baseline 8.4 Results and Analysis 8.5 Error Analysis 8.6 Annotation Anomalies 8.7 Ablation Analysis 9 Error Analysis in Individual Scaffold Task 10 Conclusion and Future Work References Automated Mining of Leaderboards for Empirical AI Research 1 Introduction 2 Related Work 3 Our Leaderboards Labeled Corpus 4 Leaderboard Extraction Task Definition 4.1 DocTAET Context Feature 5 Transformer-Based Leaderboard Extraction Models 5.1 Bert Models 5.2 XLNet 6 Automated Leaderboard Mining 6.1 Experimental Setup 6.2 Experimental Results 6.3 Integrating ORKG-TDM in Scholarly Digital Libraries 7 Conclusion and Future Work References A Prototype System for Monitoring Emotion and Sentiment Trends Towards Nuclear Energy on Twitter Using Deep Learning 1 Introduction 2 Related Work 3 Emotion Classification 3.1 Dataset Preparation 3.2 Model Building and Results 4 Aspect Based Sentiment Analysis 4.1 Dataset Preparation 4.2 Model Building and Results 5 Prototype System for Monitoring Emotion and Sentiment Trends 5.1 System Architecture and Description 5.2 Visualization 6 Conclusion References Author Index