Computational Linguistics: 16th International Conference of the Pacific Association for Computational Linguistics, PACLING 2019, Hanoi, Vietnam, ... in Computer and Information Science (1215))

Description

This book constitutes the refereed proceedings of the 16th International Conference of the Pacific Association for Computational Linguistics, PACLING 2019, held in Hanoi, Vietnam, in October 2019.  The 28 full papers and 14 short papers presented were carefully reviewed and selected from 70 submissions. The papers are organized in topical sections on text summarization; relation and word embedding; machine translation; text classification; web analyzing; question and answering, dialog analyzing; speech and emotion analyzing; parsing and segmentation; information extraction; and grammar error and plagiarism detection. Preface Organization Contents Text Summarization A Submodular Approach for Reference Recommendation 1 Introduction 2 Related Work 3 Submodularity Background 3.1 Definitions 3.2 Submodular Functions Used in Document Summarization 4 Submodular Reference Recommendation 4.1 Non-monotone Submodular Functions 4.2 Monotone Submodular Functions 5 Experiments 5.1 Corpus 5.2 Evaluation Metrics 5.3 Experimental Settings 5.4 Parameter Tuning 5.5 Performance Comparison 6 Conclusion References Split First and Then Rephrase: Hierarchical Generation for Sentence Simplification 1 Introduction 2 Split-First-and-Then-Rephrase Model 3 Experiments 3.1 Dataset 3.2 Training Details 3.3 Results 3.4 Segmentation Analysis 3.5 Error Analysis 4 Related Work 4.1 Other Approaches to the Split-and-Rephrase Task 4.2 Hierarchical Text Generation in Other Tasks 5 Conclusion References Abstractive Text Summarization Using LSTMs with Rich Features 1 Introduction 2 Related Work 3 Proposed Model 3.1 Baseline Model 3.2 Our Proposed Model 4 Experiments and Results 4.1 Dataset 4.2 Processing Data 4.3 Experiments 4.4 Results 5 Conclusions References Relation and Word Embedding SemSeq: A Regime for Training Widely-Applicable Word-Sequence Encoders 1 Introduction 2 Related Work 3 Proposed Approach 3.1 Word-Sequence Extraction 3.2 Model Architecture 4 Experiments 4.1 Training Data 4.2 Models and the Training Setup 4.3 Evaluation Setup 4.4 Word-Sequence Length Impact 5 Results and Discussion 5.1 Supervised Tasks 5.2 Unsupervised Tasks 6 Conclusion References Learning to Compose Relational Embeddings in Knowledge Graphs 1 Introduction 2 Background 2.1 Knowledge Graph Embedding Methods 2.2 Relational Walk 2.3 Inference in Knowledge Graphs 3 Relation Composition 3.1 Unsupervised Relation Composition 3.2 Supervised Relation Composition 4 Experiments 4.1 Datasets 4.2 Relation Composition Ranking 4.3 Triple Classification 5 Conclusion References Context-Guided Self-supervised Relation Embeddings 1 Introduction 2 Related Work 2.1 Pattern-Based Approach for Relations 2.2 Compositional Approach for Relations 2.3 Hybrid Approaches for Relations 3 Method 3.1 Pseudo Relation Labels 4 Experiments 4.1 Datasets 4.2 Comparison Methods 4.3 Implementation Details 4.4 Experimental Results 5 Conclusion References Evaluation of Embedded Vectors for Lexemes and Synsets Toward Expansion of Japanese WordNet 1 Introduction 2 Basic Approach to Construct Embedded Vectors Taking into Account Thesaurus Structure 2.1 AutoExtend 3 Applying AutoExtend-Based Approach to JWN 4 Experiments 4.1 Experimental Setup 4.2 Experimental Results 5 Discussions 6 Conclusions References Neural Rasch Model: How Do Word Embeddings Adjust Word Difficulty? 1 Introduction 2 Related Work 3 Previous Models 3.1 Rasch Model 3.2 Logistic Regression Models 4 Proposed Model 5 Experiments 6 Conclusions References Machine Translation Dynamic Fusion: Attentional Language Model for Neural Machine Translation 1 Introduction 2 Previous Works 2.1 Shallow Fusion 2.2 Cold Fusion 2.3 Simple Fusion 3 Dynamic Fusion 4 Experiment 5 Discussion 5.1 Quantitative Analysis 5.2 Qualitative Analysis 5.3 Influence of Language Model 5.4 Influence of Dynamic Fusion 6 Conclusion References Improving Context-Aware Neural Machine Translation with Target-Side Context 1 Introduction 2 Model Architecture 2.1 Separated Model 2.2 Shared Model 2.3 Shared Mix Model 3 Experiments 3.1 Data 3.2 Settings 3.3 Results 4 Discussion 4.1 Weight Sharing 4.2 Language Dependency 4.3 Output Examples 4.4 Convergence of Training 5 Related Works 6 Conclusion References Learning to Evaluate Neural Language Models 1 Introduction 2 Related Work 2.1 Sentence Representations 2.2 Learning the Evaluation Measure 2.3 Evaluating Language Models 3 Methodology 3.1 Language Models 3.2 Pretrained Models for Evaluation 3.3 Model-Free Embedding Evaluation 4 Results 5 Conclusion References Recommending the Workflow of Vietnamese Sign Language Translation via a Comparison of Several Classification Algorithms 1 Introduction 2 Vietnamese Sign Language 3 Materials and Methodology 3.1 Workflow of Sign Language Translation 3.2 Dataset Collection for Syntax Conversion 3.3 Machine Translation Based on Examples 3.4 Expressing Sign Language with 3D Character 4 Experiments 4.1 Data Collection 4.2 Evaluation Metrics 4.3 Experimental Results 4.4 Remarks 5 Conclusion References Text Classification Document Classification by Word Embeddings of BERT 1 Introduction 2 Related Work 3 Proposed Method 3.1 BERT 3.2 Using Word Embedding of [CLS] 3.3 Using Word Embeddings 3.4 Combined Use with BOW Models 4 Experiments 4.1 Datasets 4.2 Japanese BERT Pre-training Model 4.3 Results 5 Discussion 5.1 Feature Vector Weights from BOW Model and BERT 5.2 Comparison with Distributed Representations 5.3 Combined Use of Lower Layer Word Embeddings 5.4 Fine-Tuning 6 Conclusion References Deep Domain Adaptation for Low-Resource Cross-Lingual Text Classification Tasks 1 Introduction 2 Related Work 3 Methods 3.1 Bilingual Embedding 3.2 Parameter Sharing 3.3 Domain Adaptation and Objective Function 4 Experiments 4.1 Training Evaluation Setup 4.2 Experimental Results 5 Discussion and Conclusion References Multi-task Learning for Aspect and Polarity Recognition on Vietnamese Datasets 1 Introduction 2 Related Work 3 Our Approach 3.1 Problem Formulation 3.2 Multi-task Model 4 Datasets and Experiments 4.1 Experimental Setup 4.2 Pre-trained Word Embedding 5 Result and Discussion 6 Conclusion and Future Work References Evaluating Classification Algorithms for Recognizing Figurative Expressions in Japanese Literary Texts 1 Introduction 2 Related Work 3 Datasets and Annotation 3.1 Datasets for Classification Experiment 3.2 Annotation 4 Classification Experiment 5 Discussion and Future Work References Web Analysing Model-Driven Web Page Segmentation for Non Visual Access 1 Introduction 2 Related Works 3 Clustering Strategies 4 Reading Strategies and Seeds Positioning 5 Pre-clustering Guided Expansion 6 Quantitative Evaluation 7 Statistical Evaluation 8 Conclusions References Update Frequency and Background Corpus Selection in Dynamic TF-IDF Models for First Story Detection 1 Introduction 2 First Story Detection 3 Dynamic Term Vector Models for First Story Detection 4 Experimental Design 4.1 Target Corpus 4.2 Background Corpora 4.3 Update Frequencies 4.4 FSD Evaluation 5 Results and Analysis 5.1 Comparisons Across Different Update Frequencies 5.2 Comparisons Across Different Background Corpora 5.3 Comparisons Across Mini Corpora 6 Conclusion References A Pilot Study on Argument Simplification in Stance-Based Opinions 1 Introduction 2 Related Work 3 Background 4 Bipartite Graph-Based Opinion Matching 4.1 Unsupervised Sentence Embedding 4.2 Supervised Sentence Similarity 4.3 Sentiment and Target 5 Experiments and Results 5.1 Evaluation Measures 5.2 Analysis of Results 6 Conclusion References Automatic Approval of Online Comments with Multiple-Encoder Networks 1 Introduction 2 Problem Description 3 Proposed Model 3.1 Model Structure 3.2 Self-attention Encoder 4 Model Variants 4.1 Baselines 4.2 Encoders 4.3 Model Configurations 4.4 Pretrained Embeddings 5 Evaluation 5.1 Performance Considerations 6 Conclusion References Question and Answering, Dialog Analyzing Is the Simplest Chatbot Effective in English Writing Learning Assistance? 1 Introduction 2 Related Work 3 Method 4 Results 5 Discussion 6 Conclusions A Bi-grams Obtained from the Experimental Results References Towards Task-Oriented Dialogue in Mixed Domains 1 Introduction 2 Methodology 2.1 Sequicity 2.2 Multi-domain Dialogue State Tracking 3 Experiments 3.1 Datasets 3.2 Experimental Settings 3.3 Results 3.4 Error Analysis 4 Conclusion A Example Dialogues References Timing Prediction of Facilitating Utterance in Multi-party Conversation 1 Introduction 2 Related Work 3 Data Construction 3.1 AMI Meeting Corpus 3.2 Facilitating Utterance 4 Method 5 Experiment 5.1 Discussion About Features 5.2 Discussion About Sa 5.3 Discussion About Dialogue Act 5.4 Comparison with a Baseline 5.5 Additional Features About PM 6 Conclusion References Evaluating Co-reference Chains Based Conversation History in Conversational Question Answering 1 Introduction 2 Related Work 3 Problem Formulation 4 QANet Model for CoQA 4.1 Input Embedding Layer 4.2 Embedding Encoding Layer 4.3 Attention Layer 4.4 Model Encoding Layer 4.5 Output Layer 4.6 Inference 5 Experiments 5.1 Evaluation Metric 5.2 Implementation 5.3 Results 6 Conclusion References Speech and Emotion Analyzing Multiple Linear Regression of Combined Pronunciation Ease and Accuracy Index 1 Introduction 2 Compilation of Phonetic Learner Corpus 2.1 Collection of Pronunciation Data 2.2 Annotation of Pronunciation Data 2.3 Properties of Phonetic Learner Corpus 3 Verification of Reliability and Validity 4 Pronounceability Measurement 5 Conclusion References Rap Lyrics Generation Using Vowel GAN 1 Introduction 2 Related Works 3 Architecture 3.1 Tool 3.2 Data Set 3.3 GAN 3.4 Sequence-to-Sequence 3.5 Generation Procedure 4 Evaluation 4.1 GAN 4.2 Sequence-to-Sequence 5 Discussion 6 Conclusion References Emotion Recognition for Vietnamese Social Media Text 1 Introduction 2 Related Work 3 Corpus Construction 3.1 Process of Building the Corpus 3.2 Annotation Guidelines 3.3 Corpus Evaluation 3.4 Corpus Analysis 4 Methodology 4.1 Machine Learning Models 4.2 Deep Learning Models 5 Experiments and Error Analysis 5.1 Corpus Preparation 5.2 Experimental Settings 5.3 Experimental Results 5.4 Error Analysis 6 Conclusion and Future Work References Effects of Soft-Masking Function on Spectrogram-Based Instrument - Vocal Separation 1 Introduction 2 Related Works 3 Methodology 4 Results and Analysis 5 Conclusions and Future Works References Parsing and Segmentation Japanese Predicate Argument Structure Analysis with Pointer Networks 1 Introduction 2 Related Work 3 Japanese Predicate Argument Structure Analysis (PASA) with Sequence Labeling 3.1 Input Layer 3.2 RNN Layer 3.3 Output Layer 4 Japanese PASA with Pointer Networks 4.1 Pointer Networks 4.2 Decoding 5 Experiments 5.1 Setting 5.2 Results 6 Discussion 7 Conclusion References An Experimental Study on Constituency Parsing for Vietnamese 1 Introduction 2 Background 2.1 Constituency Parsing 2.2 Distributed Word Representations 3 Experiments 3.1 Dataset 3.2 Constituency Parsers 3.3 Evaluation Metrics 3.4 Errors Analysis 4 Conclusions References Antonyms-Synonyms Discrimination Based on Exploiting Rich Vietnamese Features 1 Introduction 2 Related Works 3 Proposed Method 3.1 Vietnamese Word-Level Patterns 3.2 Local Mutual Information 3.3 The ViASNet Architecture 4 Vietnamese Dataset Construction 4.1 Selecting Antonymous Word Pairs 4.2 Selecting Synonymous Word Pairs 5 Experiments 5.1 Baseline Models 5.2 Experimental Settings 5.3 Experimental Results 6 Conclusion References Towards a UMLS-Integratable Vietnamese Medical Terminology 1 Introduction 2 The Unified Medical Language System 3 Approaches for Term Acquisition 3.1 Corpus-Based Term Acquisition 3.2 Crowdsourcing Approach 4 Construction of a Medical Terminology for Vietnamese 4.1 Terminology Compilation from Existing Vocabularies 4.2 Corpus-Based Term Extraction 5 Conclusions References Vietnamese Word Segmentation with SVM: Ambiguity Reduction and Suffix Capture 1 Introduction 2 Our Approach 2.1 Problem Representation 2.2 Feature Extraction 3 Experiment and Result 3.1 Corpora 3.2 Experimental Setup 3.3 Feature Selection Results 3.4 Main Results 3.5 Analyses 4 Conclusion and Future Work References An Assessment of Substitute Words in the Context of Academic Writing Proposed by Pre-trained and Specific Word Embedding Models 1 Introduction 2 Specific Word Embedding Model Trained on ACL-ARC 3 Large Pre-trained Models 4 Experiments 4.1 Evaluation Using Machine Translation 4.2 Human Judgement 4.3 Discussion 5 Conclusion References Effective Approach to Joint Training of POS Tagging and Dependency Parsing Models 1 Introduction 2 Related Works 3 Methodology 3.1 Graph-Based Dependency Parsing 3.2 Encoder 3.3 Biaffine Attention Mechanism 3.4 Part-of-Speech Tagging 3.5 Objective Function of the Joint Training Model 4 Experiments 4.1 Setup 4.2 Main Results 5 Error Analysis 6 Conclusion References Information Extraction Towards Computing Inferences from English News Headlines 1 Introduction 1.1 Inferences 1.2 Related Work 1.3 Linguistic Definitions and Characteristics of Headlines 1.4 Relevance of This Work 2 Data 2.1 Format of the Data 3 Proposed Method 3.1 Extracting the Dependencies 3.2 Rule-Based System for Inference Generation 4 Results and Discussion 5 Annotation Guidelines 5.1 Purpose of Annotation 5.2 Guidelines for Annotating Presuppositions 6 Conclusions and Future Work References Extraction of Food Product and Shop Names from Blog Articles Using Named Entity Recognition 1 Introduction 2 Related Work 2.1 Named Entity Recognition 2.2 Japanese NER 2.3 NER in Noisy User-Generated Texts 3 Extraction of Food Product and Shop Names 3.1 Construction of Training Data 3.2 CRF Model 3.3 Neural Models 4 Experiment 4.1 Experimental Data 4.2 Evaluation 4.3 Model Parameters 4.4 Results 4.5 Additional Experiments and Results 5 Discussion 5.1 Quantitative Evaluation 5.2 Error Analysis 6 Conclusion References Transfer Learning for Information Extraction with Limited Data 1 Introduction 2 Related Work 3 Proposed Approach 3.1 Task Definition 3.2 Proposed Model 4 Settings and Evaluation Metric 4.1 Dataset 4.2 Baselines 4.3 Evaluation Metric 5 Results and Discussion 6 Error Analysis 7 Conclusion References Self-deprecating Humor Detection: A Machine Learning Approach 1 Introduction 2 Related Work 3 Proposed Approach 3.1 Layer-1 3.2 Layer-2 4 Feature Extraction 4.1 The Features Followed from Baseline 4.2 Newly Proposed Features 5 Datasets 6 Experiment Setup and Results 6.1 Evaluation Metrics 6.2 Evaluation Results and Comparative Analysis 7 Conclusion and Future Work References Grammar Error and Plagiarism Detection Deep Learning Approach for Vietnamese Consonant Misspell Correction 1 Introduction 1.1 Vietnamese Language 1.2 Mistyped Errors and Misspelled Errors 2 Related Works 3 Model Construction 3.1 Dataset Preparation 3.2 Misspell Direction Encoding 3.3 Deep Learning Model 3.4 Experimental Results 4 Conclusion References Grammatical Error Correction for Vietnamese Using Machine Translation 1 Introduction 2 Related Work 3 Our Method 3.1 Machine Translation 3.2 Our Method for Vietnamese Grammatical Error Correction 4 Experiments 4.1 Dataset 4.2 Settings 4.3 Results and Discussions 5 Conclustion and Future Work References Developing a Framework for a Thai Plagiarism Corpus Abstract 1 Introduction 2 Related Corpora 2.1 Existing Plagiarism Corpora 2.2 Thai Plagiarism Corpus 3 Thai Plagiarism Corpus Construction 3.1 The Framework of Thai Plagiarism Corpus Construction 3.2 The Creation of Simulated Thai Plagiarism Documents 3.3 The Creation of Artificial Thai Plagiarism Documents 4 An Example of Thai Plagiarism Cases 5 The General Statistics of Thai Plagiarism Corpus 6 The Experimental Results and Discussions 7 Conclusions and Future Works References Author Index