Artificial Intelligence in HCI: 4th International Conference, AI-HCI 2023 Held as Part of the 25th HCI International Conference, HCII 2023 Copenhagen, Denmark, July 23–28, 2023 Proceedings, Part II

Foreword
HCI International 2023 Thematic Areas and Affiliated Conferences
List of Conference Proceedings Volumes Appearing Before the Conference
Preface
4th International Conference on Artificial Intelligence in HCI (AI-HCI 2023)
HCI International 2024 Conference
Contents – Part II
Contents – Part I
Artificial Intelligence for Language, Text, and Speech-Related Tasks
Towards Modelling Elaborateness in Argumentative Dialogue Systems
	1 Introduction
	2 Related Work
		2.1 Argumentative/Mobile Dialogue Systems
		2.2 Language and Conversational Style Adaption
	3 Elaborateness Model
		3.1 System Utterance Adaptation
		3.2 Elaborateness Score
	4 System Framework and Architecture
		4.1 Dialogue Framework and Model
		4.2 Interface and NLU Framework
	5 User Study Setting
	6 Results and Discussion
		6.1 Validity (H1)
		6.2 Validity (H2)
		6.3 Validity (H3)
	7 Conclusion and Future Work
	References
AI Unreliable Answers: A Case Study on ChatGPT
	1 Introduction
	2 ChatGPT
	3 The Reliability of the Answers Provided by ChatGPT
		3.1 Creativity
		3.2 Search/Text Summarization/Translation
		3.3 Problem Solving
		3.4 Logic
	4 The User Study
		4.1 Study Planning
		4.2 Participants
		4.3 Procedure
		4.4 Results
	5 Discussion and Lesson Learned
	6 Conclusion
	References
Conversation-Driven Refinement of Knowledge Graphs: True Active Learning with Humans in the Chatbot Application Loop
	1 Introduction
	2 Problem Identification and Objectives
	3 Background
		3.1 Human-in-the-Loop (HitL) Machine Learning
		3.2 Conversational Artificial Intelligence
	4 Design and Development of the Artifact
		4.1 KG Weakness Identification
		4.2 Relevant User Identification
		4.3 Answer Validation
	5 Demonstration
		5.1 Implementation
		5.2 Case Study
	6 Related Work
	7 Discussion, Limitations, and Future Work
	References
The Relevance of Perceived Interactivity for Disclosure Towards Conversational Artificial Intelligence
	1 Introduction
		1.1 Interactivity
		1.2 Differentiation of Disclosure
		1.3 Current Research
	2 Study 1
		2.1 Method
		2.2 Results
		2.3 Discussion
	3 Study 2
		3.1 Method
		3.2 Results
		3.3 Discussion
	4 Overall Discussion
		4.1 Strengths and Limitations
		4.2 Conclusion
	References
Towards Human-Centered Design of AI Service Chatbots: Defining the Building Blocks
	1 Introduction
	2 Background
		2.1 Chatbots
		2.2 Chatbots in Customer Service
		2.3 Human-Centered Design
	3 Study Design
		3.1 Phase 1 – Designing the Review
		3.2 Phase 2 – Conducting the Review
		3.3 Phase 3 – Analysis
	4 Results
		4.1 Theme 1: Service Chatbot’s Purpose is to Serve User and Their Needs
		4.2 Theme 2: Trust is Essential for the Service Chatbot Uptake
		4.3 Theme 3: Chatbot Design Combines Dialogue Design, UI Design, and Bot Personality to Create Positive UX
	5 Discussion
		5.1 Future Work
		5.2 Limitations
	6 Conclusion
	Appendix
	References
User Experience for Artificial Intelligence Assistant: Focusing on Negative Side Effects
	1 Introduction
	2 Method
		2.1 Participants
		2.2 Prototype
		2.3 Procedure
	3 Results
	4 Discussion
	5 Conclusions
	References
CAPTAIN: An AI-Based Chatbot for Cyberbullying Prevention and Intervention
	1 Introduction
	2 Method
	3 Result
	4 Discussion and Conclusion
	References
CADNCI—Connected-Audio Digital Note-Taking for Consecutive Interpretation: Optimal Setup for Practice and Assessment
	1 Motivation
	2 The Problems
		2.1 Memory
		2.2 Pen and Paper
		2.3 The Note-Taking Action
	3 Digital Note-Taking
		3.1 The Literature
		3.2 The Digital Generation
		3.3 Quest for the Ideal Setup
	4 Same-Language Consecutive Practice
	5 The CADNCI System
		5.1 Demonstrations
		5.2 System Advantages
	6 Online Group Practice Scenarios
		6.1 Scenario 1: Allowing Everyone to Record Their Note-Taking Action and Rendition—Each Person in Their Own Breakout Room
		6.2 Scenario 2: Multiple People in One Breakout Room, Each Playing a Different Role
	7 Conclusion
	References
Misrecognized Utterance Identification in Support Systems for Daily Human-to-Human Conversations
	1 Introduction
	2 Daily Conversation Database
		2.1 Conversation Recording
	3 Recognition Error Characteristics for Daily Conversation
		3.1 Extraction Rates of Fundamental Frequency (F0)
		3.2 Number of Words for Each Utterance
	4 Conclusion
	References
BLEU Skies for Endangered Language Revitalization: Lemko Rusyn and Ukrainian Neural AI Translation Accuracy Soars
	1 Introduction
		1.1 The Problem
		1.2 System Under Study
	2 State of the Art
	3 Materials and Methods
		3.1 Materials
		3.2 Methods
	4 Results and Discussion
		4.1 English to Lemko Translation Quality
		4.2 Lemko to English Translation
	5 Conclusion
	References
Translators as Information Seekers: Strategies and Novel Techniques
	1 Introduction
	2 Information Seeking Strategies and Examples
		2.1 The Objectives of Information Seeking
		2.2 Strategy 1: Top-Ranked Entries and Featured Snippets Returned by Search Engines Based on Bottom-Up Search Strategy
		2.3 Strategy 2: Guesstimate Among Alternative Answers Based on the Number of Results Returned by Customized Keyword Search Using a Large-Scale Web Search Engine
		2.4 Strategy 3: New Search Experience Enabled by Long-Form Question and Answer Conversational AI (ChatGPT)
		2.5 Strategy 4: Conversational Search Engine Combining Large Language Model and Large-Scale Web Search Engines
	3 Information Seeking Examples
		3.1 Example 1: Medical Term (Medical Record)
		3.2 Example 2: Technology Terms (Patent Claim)
		3.3 Example 3: Ambiguous Source Expression (Social Media Blog Post)
	4 Conclusion
	References
Comparing Sentiment Analysis and Emotion Analysis of Algorithms vs. People
	1 Background
		1.1 History of Sentiment Analysis
	2 Data Gathering
		2.1 First Survey
		2.2 Second Survey
		2.3 Third Survey
	3 Text Pre-processing
	4 Sentiment Analysis
		4.1 Sentiment Sources
		4.2 Algorithms
		4.3 Results
	5 Emotion Analysis
		5.1 Emotion Sources
		5.2 Results
	6 Conclusions
	References
Tell Me, What Are You Most Afraid Of? Exploring the Effects of Agent Representation on Information Disclosure in Human-Chatbot Interaction
	1 Introduction and Related Work
		1.1 Chatbot Visual Appearance and Classification
		1.2 Chatbots in Health Care and Therapy
		1.3 Information Disclosure
		1.4 Chatbots as Social Actors
	2 Theoretical Framework and Hypothesis Development
	3 Methodology
	4 Instruments
		4.1 Questions
		4.2 OID Analysis Scheme
	5 Results
	6 Discussion of Results and Their Limitations
	7 Conclusion and Future Outlook
	References
Human Decision-Making and Machine Assistance in Subtitle Translation Through the Lens of Viewer Experience
	1 Introduction
	2 First Cornerstone: Accurate Translation
		2.1 Three-Dimensional Context
		2.2 Tricky Translations
	3 Second Cornerstone: Scriptwriting Mindset
		3.1 Creative Recasting
		3.2 Subtitle Flow
	4 Third Cornerstone: Effortless Reading
		4.1 Concise Writing
		4.2 Shot Changes
	5 Fourth Cornerstone: Equivalent Experience
		5.1 Subtitle Segmentation
		5.2 Creative Intent
	6 Conclusion
	References
Human-AI Collaboration
Human Satisfaction in Ad Hoc Human-Agent Teams
	1 Introduction
	2 Related Work
	3 Research Hypotheses
	4 Methodology
		4.1 Experimentation Domain Characteristics and Description
		4.2 Interaction Protocols
		4.3 Experimental Setup
		4.4 Satisfaction and Task Likeability Measurements
	5 Results
	6 Discussion and Future Work
	References
Composite Emotion Recognition and Feedback of Social Assistive Robot for Elderly People
	1 Introduction
	2 DarumaTO Social Robot
	3 Design of Composite Emotion Recognition
		3.1 Voice Emotion
		3.2 Semantic Emotion
		3.3 Topic Emotion
		3.4 Integration
	4 Emotion Feedback with Facial Expression
	5 Experiments and Analysis
		5.1 Voice Emotion
		5.2 Semantic Emotion
	6 Conclusion and Future Work
	References
My Actions Speak Louder Than Your Words: When User Behavior Predicts Their Beliefs About Agents\' Attributes
	1 Introduction
	2 Data
		2.1 Study 1 Task and Treatment Conditions
		2.2 Study 2 Task and Treatment Conditions
		2.3 Measures
	3 Analyses
	4 Discussion
		4.1 Limitations
	5 Conclusions and Future Work
	References
Collaborative Appropriation of AI in the Context of Interacting with AI
	1 Introduction
	2 The Case
	3 Method
	4 Findings
		4.1 Problems and Proposals
		4.2 The Customer Service Team
		4.3 Further Development of the M&A-Platform
	5 Discussion and Conclusion
	References
Understanding User Experience with AI-Assisted Writing Service
	1 Introduction
	2 Literature Review
		2.1 AI-Assisted Writing Service
		2.2 AI-Assisted Service User Experience
	3 Methods
		3.1 Participants
		3.2 Prototype
		3.3 Procedures
		3.4 Measures
	4 Results
		4.1 User Experience According to Creativity Level
		4.2 User Experience According to Writing Case
	5 Discussions
		5.1 Significant Findings
		5.2 Limitations
	6 Conclusions
	References
A Methodology for Personalized Dialogues Between Social Robots and Users Based on Social Media
	1 Introduction
	2 Related Works
	3 Our Methodology for Human-Robot Dialogue Generation
	4 Use Case
	5 Conclusions and Future Works
	References
EEG-Based Machine Learning Models for Emotion Recognition in HRI
	1 Introduction
	2 Materials and Methods
		2.1 Valence and Arousal
		2.2 Classifiers
	3 Experiment
		3.1 Partecipants
		3.2 Stimuli
		3.3 Acquisition Device
		3.4 Dataset
		3.5 Feature Extraction
		3.6 Feature Selection
		3.7 Optimization Workflow Procedure
	4 Results Analysis
	5 Conclusions and Discussion
	References
We Want AI to Help Us
	1 Introduction
	2 Related Work
		2.1 Working as a Reactor Operator
		2.2 The User-Centered Design Approach in NPPs
		2.3 AI in NPPs
	3 Methodology and Methods
		3.1 Data Gathering and Data Analysis
	4 Results
		4.1 Understanding of AI
		4.2 Pain Points of Operator’s Work
		4.3 What AI Can Do for Operators?
	5 Discussion
	6 Conclusion
	References
The Design of Transparency Communication for Human-Multirobot Teams
	1 Introduction
	2 Empirical Testbed
	3 Team Decision-Making
	4 Team Explanation
		4.1 Automatic Explanation of Joint Assessment
		4.2 Automatic Explanation of Team Activities
	5 Conclusion
	References
User Transparency of Artificial Intelligence and Digital Twins in Production – Research on Lead Applications and the Transfer to Industry
	1 Introduction
	2 Goal of the Research Project
	3 Predictive Maintenance Through Anomaly Detection
	4 Predictive Quality by Optimizing Process Parameters Using Deep Reinforcement Learning
		4.1 Deep Reinforcement Learning
		4.2 Offline Reinforcement Learning
	5 Explainable AI, Digital Twin and Augmented Reality
		5.1 Explainability
	6 Outlook
	References
Artificial Intelligence for Decision-Support and Perception Analysis
Redundancy in Multi-source Information and Its Impact on Uncertainty
	1 Introduction
	2 Discussion
		2.1 Shannon Entropy at the Sensor and Sensor System Levels
		2.2 The Concept of Uncertainty of Information and Identifying Categories
		2.3 The Value of Redundancy in Uncertain Information
		2.4 A Two-Dimensional Model of Redundancy and Entropy
	3 Conclusion and Future Work
	References
Experimental Validation of a Multi-objective Planning Decision Support System for Ship Routing Under Time Stress
	1 Introduction
	2 Design of a Decision-Support System for Ship Routing
		2.1 Recent Advancements in AI-Assisted Naval Planning
		2.2 DSS Design Principles
		2.3 Levels of Decision Support
	3 Experimental Design for the Evaluation of DSS Levels
		3.1 Scorecard Routes and Difficulty Levels
		3.2 Participants
		3.3 Procedure
	4 Follow-Up Experiment Involving Time Pressure
	5 Results and Discussion
		5.1 Which DSS is Better?
		5.2 Time Pressure Findings
		5.3 Indirect Inference of Human Preferences
	6 Limitations and Future Work
	7 Conclusion
	References
Modeling Users\' Localized Preferences for More Effective News Recommendation
	1 Introduction
	2 Related Work
	3 Experimental Methodology
		3.1 Dataset
		3.2 Baseline Methods and Evaluation Metrics
	4 Experimental Results and Analysis
		4.1 Experimental Design
		4.2 Main Results and Discussion
	5 Conclusions
	References
What Color is Your Swan? Uncertainty of Information Across Data
	1 Introduction
	2 Use Case
	3 Conclusion
	References
AI Enabled Decision Support Framework
	1 Introduction
	2 Enhanced Tactical Inferencing (ETI)
		2.1 Sentry Agents Framework (SAGE)
		2.2 ETI Agents
	3 Uncertainty of Information (UoI)
	4 Integration of ETI with Tactical Service-Oriented Architecture
	5 Conclusion
	References
A Framework for Contextual Recommendations Using Instance Segmentation
	1 Introduction
	2 Related Work
	3 Methodology
		3.1 Fully-Convolutional Instance-Aware Semantic Segmentation (FCIS)
		3.2 Mask-Region-Based Convolutional Neural Network (Mask-RCNN)
		3.3 RetinaMask
		3.4 Path Aggregation Network (PA-Net)
		3.5 Mask Scoring Region-based Convolutional Neural Network (MS-RCNN)
		3.6 You only Look at Coefficients (YOLACT)
		3.7 Model Selection
	4 Experimental Results
		4.1 Quantitative Data
		4.2 Qualitative Data
	5 Conclusions
	References
Development of a Domain Specific Sentiment Lexicon (DSSL) for Evaluating the Quality of Experience (QoE) of Cloud Gaming
	1 Introduction
	2 Related Work
		2.1 Approach of Evaluating the CGQoE
		2.2 EDL Approach Based on the DSSL, Semantic Similarity, and Word Frequency
	3 Experiment
		3.1 Dataset About Cloud Gaming
		3.2 Creation of Cloud Gaming QoE DSSL
		3.3 Filtration of Dataset Based on the CGQoE DSSL
		3.4 Emotion Distribution Learning of UGC Based on DSSL
		3.5 Correction of Dataset Based on DSSL Labeling Results
		3.6 Result
	4 Discussion and Future Work
	References
Is Turn-Shift Distinguishable with Synchrony?
	1 Introduction
	2 Related Works
	3 Corpus
	4 Analysis
	5 Turn-shift Classification Models
	6 Conclusion and Discussion
	References
Innovations in AI-Enabled Systems
I-Brow: Hierarchical and Multimodal Transformer Model for Eyebrows Animation Synthesis
	1 Introduction
	2 Background and Related Work
		2.1 Gesture Generation Models
		2.2 Facial Gestures Synthesis Models
	3 Multimodal Data Features
	4 Training and Testing Dataset
	5 ``I-Brow\" Model for Speech-driven Upper Facial Gesture Generation
		5.1 Word Level Model Architecture
		5.2 IPU-level Model Architecture - ``I-Brow\"
		5.3 Training and Testing Procedures
	6 Evaluation Measures
		6.1 Objective Measures
		6.2 Subjective Measures
	7 Objective Evaluation Results
		7.1 ``I-Brow\" Model
		7.2 Baseline Models
	8 Perceptual Evaluation Results
	9 Discussion
	10 Conclusions and Future Work
	References
YOLO NFPEM: A More Accurate Iris Detector
	1 Introduction
	2 Related Works
		2.1 Small Object Detection
		2.2 Iris Detection
	3 Methods
		3.1 Network Backbone
		3.2 Small Size Feature EnhancementSMALL
	4 Multiscale Eye Dataset
	5 Results
	6 Conclusion
	References
Detecting Scoreboard Updates to Increase the Accuracy of ML Automatic Extraction of Highlights in Badminton Games
	1 Research Background
	2 Sports Highlight Models
		2.1 YOLO Model
		2.2 Research on Sports Object Detection Models
	3 Methodology
		3.1 Data Collection Method
		3.2 Data Collection Method
	4 Methodology
		4.1 Data Collection Method
		4.2 Data Collection Method
		4.3 Model Building Procedures
		4.4 Model Adoption and Algorithm
		4.5 Model Evaluation Method
	5 Results
	6 Discussions and Conclusion
	References
A Drone-mounted Depth Camera-based Motion Capture System for Sports Performance Analysis
	1 Introduction
	2 Related Work
	3 Our Prototype System
		3.1 Drone Hardware
		3.2 Depth Camera
		3.3 Onboard Computation
		3.4 Software
	4 Application Areas and Initial Tests
		4.1 Use-case Evaluation
		4.2 Skeleton Tracking Accuracy
		4.3 Autonomous Flying
	5 Challenges
		5.1 Power and Weight
		5.2 Computation, Storage, and Communications Capacity
		5.3 Cameras and Optics
		5.4 Skeleton Tracking Lack of Accuracy
		5.5 Autonomous Flight
		5.6 Time Synchronization with Other Devices
	6 Conclusions
	References
General Agent Theory of Mind: Preliminary Investigations and Vision
	1 Introduction
	2 Leveraging Existing Ideas to Re-create ToMnet Experiments
	3 Work Thus Far
		3.1 Simulation and Data Generation
		3.2 Model Development
	4 Scaling and Potential Applications
	5 Conclusion
	References
Trust Crisis or Algorithm Boost? An Investigation of Artificial Intelligence Divination Applications in China
	1 Introduction
	2 Digital Divination and Religious Culture
	3 AI Algorithm and Interactive Divination Application Platform
	4 Research Methods and Participants
	5 Digital Practice of AI Divination
		5.1 Algorithm Boost in Online Environment
		5.2 Trust Crisis in Offline Environment
		5.3 Human or Non-human: Towards AGI
	6 Conclusion and Discussion
	Appendix
	References
Human-Machine Learning Approach for Ejection System Design in Injection Molds
	1 Introduction
	2 Previous Work
	3 Concept
	4 The Machine Model for Positioning Ejection Elements
		4.1 Process for Deriving a Positioning Model from Human Knowledge
		4.2 The Positioning Model for Ejection Pins
	5 The Feedback System
	6 The Training Process
	7 Validation
	8 Discussion
	9 Outlook and Further Research
	References
A Weightlifting Clean and Jerk Team Formation Model by Considering Barbell Trajectory and LSTM Neural Network
	1 Introduction
	2 Materials and Methods
		2.1 Motion Phase of Clean and Jerk
		2.2 Experimental Procedure
		2.3 Data Collection and Processing
		2.4 Neural Network Training and Parameter Adjustment
	3 Result
	4 Discussion
	5 Conclusion
	References
Incorporating the Dynamics of Climate Change into the Deep Dive Virtual Reality Underwater Site Prediction System
	1 Introduction
	2 The Multi-agent Planning Framework for the Deep Dive Simulation Component
		2.1 The Prototype Herd Movement Model
	3 Using Cultural Algorithms to Generate Optimal Caribou Migration Paths
		3.1 Original and Esker Update CA Optimized Runs Comparison
	4 Optimization Results
	5 Conclusions
	References
Using Machine Learning to Model Potential Users with Health Risk Concerns Regarding Microchip Implants
	1 Introduction
	2 Literature Review
	3 Data Collection
		3.1 Methods
		3.2 Survey Design
		3.3 Participants
	4 Data Preparation
		4.1 Thematic Coding
		4.2 Data Cleansing
		4.3 Data Imputation
		4.4 Feature Engineering
	5 Experimental Details
		5.1 Machine Learning Techniques
		5.2 Experimental Procedures
	6 Experimental Findings
		6.1 User Characteristics Based on Applied Technique
		6.2 Model Testing
	7 Conclusion and Future Works
	8 Limitation
	References
Measuring Human Perception and Negative Elements of Public Space Quality Using Deep Learning: A Case Study of Area Within the Inner Road of Tianjin City
	1 Introduction
	2 Related Work
	3 Approach
		3.1 Acquisition of Tianjin Street View Data
		3.2 MIT Place Pulse Data
		3.3 Scene Perception Based on Convolutional Neural Network
		3.4 Scene Perception Based on Convolutional Neural Network
		3.5 Correlation Analysis Between Perception and Physical Elements
	4 Experiment Results
	5 Discussion
		5.1 Measures to Improve the Experimental Process
		5.2 Countermeasures to Improve the Six Indexes
	6 Conclusion
	References
Multi-monitor System for Adaptive Image Saliency Detection Based on Attentive Mechanisms
	1 Introduction
	2 Related Works
	3 The Selective Attention Mechanism
		3.1 Our Selective Attentional Model
	4 The Multi-monitor System
	5 Conclusions and Discussions
	References
Author Index