New Frontiers in Artificial Intelligence: JSAI-isAI 2022 Workshop, JURISIN 2022, and JSAI 2022 International Session, Kyoto, Japan, June 12–17, 2022, Revised Selected Papers

Description

This book constitutes extended, revised, and selected papers from the JSAI annual conference, JSAI 2022, and the 14th International Symposium on Artificial Intelligence, JSAI-isAI 2022, held in Kyoto, Japan, in June 2022.  The 18 full papers were carefully selected from 67 submissions and presented during the two events: 16th International Workshop on Juris-informatics, JURISIN 2022, and JSAI 2022 Intenational Session. This papers present discussion on fundamental and practical issues in Juris-informatics among researchers from various backgrounds such as law, social science, information and intelligent technology, logic, and philosophy, including the conventional AI and Law area.  Preface Organization Contents JURISIN 2022 Juris-Informatics (JURISIN) 2022 Differential-Aware Transformer for Partially Amended Sentence Translation 1 Introduction 2 Task Definition 2.1 Partial Amendments in Japanese Legislation 2.2 Differential Translation for Partial Amendments 3 Related Work 3.1 Translation Methods 3.2 Evaluation Criteria 3.3 Corpora 4 Proposed Architecture 5 Experiment 5.1 Outline 5.2 Results 6 Discussion 6.1 Using Pre-amendment Sentences in Naive Transformer 6.2 Training Iterations 6.3 Using a Large-Scaled Bilingual Corpus 6.4 Investigation into n-Best Candidates 7 Summary References On Complexity and Generality of Contrary Prioritized Defeasible Theory 1 Introduction 2 Backgrounds 2.1 Normal Logic Program 2.2 Defeasible Logic Program 2.3 Ambiguity Blocking and Ambiguity Propagation 3 Contrary Prioritization 3.1 Complexity 3.2 Generality in Translation to and from Stratified Logic Program 4 Discussion 5 Conclusion References Mapping Similar Provisions Between Japanese and Foreign Laws 1 Introduction 2 Similar Document Search 2.1 Document Unit 2.2 Similarity of Bag of Words 2.3 Similarity of Vectors 3 Proposed Method 4 Experiments 4.1 Purpose 4.2 Procedure of Experiments 4.3 Creating a Correct Dataset: Japanese Laws 4.4 Creating a Correct Dataset: Foreign Laws 4.5 Evaluation Method 5 Experimental Results 5.1 Experiment 1: Japanese Laws in Japanese 5.2 Experiment 2: Japanese Laws in English 5.3 Experiment 3: Japanese Civil Code and German Civil Code in English 6 Conclusion References COLIEE 2022 Summary: Methods for Legal Document Retrieval and Entailment 1 Introduction 2 Task 1 - Case Law Information Retrieval 2.1 Task Definition 2.2 Dataset 2.3 Approaches 2.4 Results 3 Task 2 - Case Law Entailment 3.1 Task Definition 3.2 Dataset 3.3 Approaches 3.4 Results 4 Task 3 - Statute Law Retrieval 4.1 Task Definition 4.2 Dataset 4.3 Approaches 4.4 Results 5 Task 4 - Statute Law Entailment 5.1 Task Definition 5.2 Dataset 5.3 Approaches 5.4 Results 6 Conclusion References JNLP Team: Deep Learning Approaches for Tackling Long and Ambiguous Legal Documents in COLIEE 2022 1 Introduction 2 Related Work 2.1 Case Law 2.2 Statue Law 3 Approaches 3.1 Case Law 3.2 Task2 3.3 Task 3 3.4 Task 4. Statute Law Entailment and Question Answering 4 Experiments and Results 4.1 Task 1 4.2 Task2 4.3 Task 3 4.4 Task 4 5 Conclusion References Semantic-Based Classification of Relevant Case Law 1 Introduction 2 Literature Review 3 Our Method 3.1 Dataset Analysis 3.2 Details of the Approach 4 Results 4.1 Error Analysis 5 Final Remarks References nigam@COLIEE-22: Legal Case Retrieval and Entailment Using Cascading of Lexical and Semantic-Based Models 1 Introduction 2 The Task 2.1 Task 1: Legal Case Retrieval 2.2 Task 2: Legal Case Entailment 3 Data Corpus 3.1 Preprocessing 3.2 Evaluation Metrics 4 Our Methods 4.1 BM25 4.2 Sent2Vec 4.3 Sentence-BERT 4.4 Reduced-Space 4.5 Reasons for Max-Pooling 5 Results and Analysis 5.1 Task-1 5.2 Task-2 6 Conclusions References HUKB at the COLIEE 2022 Statute Law Task 1 Introduction 2 Tools and Settings 2.1 Keyword-Based IR System 2.2 BERT-Based IR and Entailment System 3 Submitted Runs and Evaluation Results 3.1 Task 3 3.2 Task 4 4 Additional Experiments 5 Summary References Using Textbook Knowledge for Statute Retrieval and Entailment Classification 1 Introduction 2 Related Work 3 Extracting Knowledge About Statutes from Textbooks 4 Statute Retrieval Task 5 Statute Entailment Classification Task 6 Evaluation 6.1 Statute Retrieval Task 6.2 Statute Entailment Classification Task 7 Conclusion and Future Work References Legal Textual Entailment Using Ensemble of Rule-Based and BERT-Based Method with Data Augmentation by Related Article Generation 1 Introduction 2 Previous Works 3 System 3.1 System Overview: Rule-Based Part and BERT-Based Part 3.2 Data Augmentation 3.3 Data Selection 3.4 Person Name Inference 3.5 Ensemble of Rule-Based Part and BERT-Based Part 4 Result and Experiments 4.1 Formal Run Results 4.2 Experiments on Training Dataset 5 Discussion 6 Conclusion and Future Works References Less is Better: Constructing Legal Question Answering System by Weighing Longest Common Subsequence of Disjunctive Union Text 1 Introduction 2 Competition Data and Task Description 3 Data Preprocessing 4 The Result 5 Discussion and Conclusion References JSAI 2022 International Session JSAI 2022 International Sessions Proposal for Turning Point Detection Method Using Financial Text and Transformer 1 Introduction 2 Related Works 3 Method 3.1 Polarity Classification Method Using Bidirectional Encoder Representations from Transformers (BERT) 3.2 Transformer+Time2Vec 3.3 Hotelling's T2 Method 4 Experiments 5 Results 6 Discussion 7 Summary Appendix A Evaluation Metrics for Time Series Deep Learning Models Appendix B Empirical Results for Industry-Specific Polarity Index References Product Portfolio Optimization for LTV Maximization 1 Introduction 2 Methods 2.1 How to Calculate LTV 2.2 Markowitz's Mean-Variance Model 2.3 Dataset 2.4 Algorithm 3 Results 3.1 Extraction of Target Customer Groups 3.2 Performance Evaluation of Rebalancing 4 Discussion 5 Conclusions References An Examination of Eating Experiences in Relation to Psychological States, Loneliness, and Depression Using BERT 1 Introduction 2 Related Work 3 Data Collection 3.1 Purpose 3.2 Method 3.3 Result and Discussion 4 Experiments 4.1 Experiment 1 4.2 Experiment 2 4.3 Experiment 3 5 General Discussion 6 Conclusion References Objective Detection of High-Risk Tackle in Rugby by Combination of Pose Estimation and Machine Learning 1 Introduction 2 Related Works 3 Dataset 3.1 High Risk Tackle Identification 3.2 Low Risk Tackle Identification 3.3 Frame Selection and Pose Estimation 3.4 Tackler and Ball Carrier Identification 4 Training and Experiments 4.1 Train/Test Split 4.2 Initial Selection of Machine Learning Models with Parameter Tuning 4.3 Final Selection of the Model with Data Augmentation 4.4 Comparison with Human Labeled Methods 5 Results 5.1 Initial Selection of Machine Learning Models 5.2 Final Selection of the Model with Data Augmentation 5.3 Comparison with Human Labeled Methods 6 Discussion and Conclusion 7 Limitation 8 Future Directions References Incremental Informational Value of Floorplans for Rent Price Prediction*6pt 1 Introduction 2 Background 2.1 Price Structure of Real Estate 2.2 Machine Learning Methods 3 Methodology 3.1 Data 3.2 Neural Network Architecture 3.3 Multiple Linear Regression 4 Results 5 Discussion 5.1 Discussion and Critique of the Neural Network 5.2 General Discussion 6 Conclusion A Appendix References Transaction Prediction by Using Graph Neural Network and Textual Industry Information 1 Introduction 2 Related Work 3 Proposed Methods 3.1 Node Features 3.2 Edge Features 3.3 Graph Attention Network 3.4 Edge Weight-Enhanced Attention Mechanism for Banking Transactions 3.5 Graph Isomorphism Network 3.6 Injectivity and Motivation to Use Textual Information 3.7 Non-probabilistic Graph Autoencoder 4 Evaluation 4.1 Dataset 4.2 Task 4.3 Baseline Methods 4.4 Experiment 5 Discussion 6 Conclusion References Overfitting Problem in the Approximate Bayesian Computation Method Based on Maxima Weighted Isolation Kernel 1 Introduction 2 Related Work 3 Background 3.1 Kernel Mean Embedding 3.2 Kernel ABC 3.3 Isolation Kernel Based on Voronoi Diagram 4 Overfitting Problem in Maxima Weighted iKernel ABC 4.1 Maxima Weighted iKernel Mapping 4.2 Meta-sampling Algorithm Based on Maxima Weighted iKernel 4.3 Weaknesses of the Algorithm and the Overfitting Problem 5 Experiments 5.1 Synthetic Data 5.2 Cancer Cell Evolution 6 Conclusion References Author Index