This study presents a Retrieval-Augmented Generation (RAG)-assisted workflow using AnythingLLM to support evidence- informed accessibility redesign. Using QR-Code interaction for older adults as a case example, the workflow retrieves accessibility evidence, maps it to redesign decisions, and formalizes the results into state-based pseudo-code. The contribution lies in a traceable design-support process that connects literature-based evidence, interface decisions, and interaction logic, rather than proposing new accessibility principles.

This study examines the visible thematic patterns of artificial intelligence (AI)–related research in Mainland China, Taiwan, Hong Kong, and Macao from 2021 to 2025. The analysis is based on Scopus-indexed journal articles and uses keyword frequency counts, VOSviewer density visualizations, and KH Coder co-occurrence networks. These methods are applied to describe how AI-related keywords appear across the four regions and how their distributions change during the five-year period. Across all datasets, terms such as machine learning, deep learning, and neural network appear frequently and occupy central positions in the visual outputs. The VOSviewer heatmaps show that regions with larger publication volumes display wider areas of keyword density, while regions with smaller datasets present more compact clusters. Beginning in 2024 and 2025, generative AI–related terms, including large language model and ChatGPT, become visible across all regions. The KH Coder networks illustrate that the four regions contain multiple clusters of co-occurring keywords, with differences in cluster size and distribution reflecting the underlying dataset scale and the topics present in each regional corpus. Overall, the results offer a descriptive account of how AI-related terms appear in the collected datasets and how their visible distributions vary among the four regions during the study period. The findings are intended to summarize observable patterns without inferring causal explanations or evaluating the significance of regional differences.

This study designs an interactive diary system that uses an artificial intelligence agent (AI Agent) and semantic generative analysis to help users break free of habitual narratives. The system will detect repetitive language and themes in the diary and provide empathetic and creative feedback about the guidance at the appropriate time, guiding users to reflect on their daily life experiences from different perspectives. Through response and learning mechanisms, AI can gradually adjust its feedback style, promote psychological resilience and life awareness, and establish a long-term interactive relationship with the AI Agent as a digital partner. The proof-of-concept prototype was implemented using prompt engineering. The results show that the empathic feedback generated using AI can help users perceive their lifestyle more positively and motivate them to make changes.

The relationship between emotions and dietary habits is complex. To enhance the dining experience, this study proposes an integrated platform that collects subjective emotional responses to meals, existing culinary components, basic customer demographics, and Google reviews. The platform uses RAG techniques to analyze the impact of various ingredients and cooking methods on specific emotions. Based on these insights, it generates innovative emotion-based menu design guidelines for restaurant reference, ultimately improving customer satisfaction and loyalty.

This study offers a thorough investigation into shared electric bicycles (E-bikes) and their role as a key element in advancing sustainable transportation within smart cities. Previous studies have predominantly employed the Technology Acceptance Model (TAM) or the Theory of Planned Behavior (TPB) to investigate user acceptance, yet few have integrated both frameworks to explore how environmental objectives-such as net-zero emissions (NZE)-influence behavioral intention in smart cities. To address this limitation, the present study incorporates key variables including subjective norms (SN), perceived usefulness (PU), user attitudes (ATU), and perceived behavioral control (PBC), while introducing NZE as a factor in user behavior analysis. Data were collected from 298 urban residents aged 18-54 via an online questionnaire. Structural equation modeling analysis was conducted to examine the relationships among SN, PU, ATU, PBC, and NZE. The results showed that SN significantly predicted PU, which confirm the strong influence of social factors on user perceptions. PU stands as a mediating variable, and it significantly influenced users' ATU and NZE, which reflected that users highly value the convenience and environmental benefits of shared e-bikes. In addition, PU is a key factor in users' understanding and adoption of NZE, demonstrating that a well-designed shared e-bike system can enhance environmental sensitivity and awareness of green behavior. Furthermore, NZE exerted a significant positive impact on PBC, indicating that greater environmental awareness enhances users' sense of behavioral control, thereby actively motivating them to opt for shared e-bikes. These findings provides empirical evidence and strategic directions for policymakers and urban planners: by strengthening social advocacy, optimizing system usefulness design, and guiding environmental awareness through green policies, cities can more effectively encourage the adoption of shared electric bicycles in smart urban environments.

Mindful eating fosters awareness of food choices and behaviors, promoting healthier dietary habits and reducing impulsive eating. This paper proposes an AI-driven mindful eating framework that combines Generative AI and mobile health technology to offer personalized dietary guidance and emotional awareness training. The AI analyzes eating behaviors, emotional states, and triggers to provide real-time feedback and adaptive recommendations. This framework promotes habit formation and sustainable mindful eating practices by incorporating interactive learning and self-guided insights. Experts in nutrition and psychology affirmed the framework’s relevance and highlighted the need for simplified input and eating context tracking. Key challenges include user engagement, data privacy, and AI transparency. Future research will focus on enhancing personalization and evaluating scalability for digital mindful eating interventions.

Recent advancements in Generative AI (GenAI) and Large Language Models (LLMs) have reshaped the potential of AI in education. This study introduces an AI-Mediated Communication (AI-MC) framework designed to enhance mentor-mentee interactions through adaptive AI-driven conversation structuring, content recommendations, and linguistic feedback. A mixed-methods study was conducted using Self-Determination Theory (SDT) and Design Thinking, combining quantitative user surveys (N = 33) and qualitative interviews. Key findings show that the "Suggesting Content using Conversational Context" feature significantly enhanced intrinsic motivation (r = .389, p = .025) and perceived competence (r = .458, p = .007), supporting users' confidence and autonomy in communication. The platform effectively supported competence and autonomy, although relatedness was rated moderately, reflecting the limitations of text-only interaction. Qualitative data reinforced that AI-MC features facilitated clearer communication and more informed mentor selection. These findings underscore the potential of AI-MC as an adaptive communication tool for mentoring, with applications that could extend to intelligent tutoring, professional coaching, and online learning platforms. Future research should aim to integrate real mentors, utilize multimodal AI communication, and offer long-term relationship support to enhance both ecological validity and relational depth further.

Artificial Intelligence-Generated Content (AIGC) is rapidly transforming various industries, including interior design. This research explores the integration of AIGC through qualitative and quantitative methods, notably text analysis based on insights from designer interviews and the Analytic Hierarchy Process (AHP) to evaluate the research results. Findings suggest that AIGC significantly enhances communication, innovation, and decision-making efficiency. However, accuracy, reliability, and human-AI interaction remain significant concerns. The study provides practical implications for designers and highlights pathways for more effective human-centered AI integration, setting foundations for broader digital transformation strategies in creative industries.

Family structures and changes in the social environment have presented parents with numerous challenges. Therefore, the study designs an integrated service platform that combines clinics, pharmacies, and user-end information and data. Caregivers can record infants and young children’s daily care and growth conditions on the platform, which then uses AI technology to analyze the children’s growth status and preferences, offering product recommendations. When infants exhibit minor symptoms, parents can seek suggestions from professional pharmacists who will record the assistance details and symptoms on the platform, sharing information to expedite doctor consultations to minimize the risk of reinfection. Besides, the clinics provide parenting columns and real-time online customer service on the platform, enabling parents to seek professional advice immediately when facing childcare issues.

This study examines how AI-Generated Content (AIGC) applications transform influencer marketing operational modes through text analysis. By comparing traditional influencer marketing with AIGC, the performance of three influencers was analyzed in terms of efficiency, interactivity, authenticity, and creativity. Results show that AIGC enhances efficiency and interactivity but faces challenges in authenticity and creativity. The study identifies key trends in influencer marketing transformation using high-frequency word analysis and co-occurrence network analyses. This study's results suggested refining marketing strategies, offering insights for future influencer marketing research and practical applications.

As discussions surrounding artificial intelligence (AI) and speculative design in education continue to grow, this study aims to assess how large language models (LLMs) analyze the application of AI-generated content (AIGC) and speculative design in current educational contexts, as well as their potential for future development. A sample of 30 articles on the application of AIGC and speculative design in education was collected, and a comprehensive text analysis method was employed, utilizing LLMs for summary analysis and opinion discussions. Through a cross-comparison of these methods, this research examines the similarities and differences in the impact of AI-driven tools on education. It explores their potential expanded functionalities in the educational field. In summary, this study integrates qualitative and quantitative data for comparative analysis, providing theoretical foundations and practical guidance on the future application of LLMs in education. The study also offers solutions to challenges and areas of concern for AIGC in education, as identified through LLM-generated inquiries, which can serve as a reference for future improvements. Further research can validate the effectiveness of these tools and explore their potential applications in diverse educational environments.

By using big data, we evaluated the user experience (UX) of cultural curation for a case study of the 2023 Hakka Expo in Taiwan (2023 HEIT). Attendees experienced a 360-degree virtual recreation of the curation, facilitated by advanced computing and processing technologies, including hardware and software that supported the immersive experience. They were interviewed to recall and compare their physical and digital UX. Responses to the User Experience Questionnaire (UEQ) were included in the quantitative big dataset. The data included interview transcripts, news articles, curatorial design proposals, and other documentation relating to the 2023 HEIT, all of which were processed using sophisticated software for text analysis. The data analysis results revealed users rated the pragmatic and hedonic aspects of 2023 HEIT on the UEQ. However, text analysis on the thick data was used to identify 11 high-frequency themes/concepts designed by curators, and only 4 aligned with the data sources. Specifically, the user interview results surfaced only 4 of these 11 core curatorial concepts from official materials. Key elements were not fully conveyed to users. This mixed-methods approach highlighted the complementary value of integrating big and thick data, along with advanced computing and processing tools, for nuanced UX evaluation. Future research is necessary to explore synergies between quantitative and qualitative analysis in assessing cultural curation and experiences.

This study investigates collaboration issues between UI designers and engineers and the application of Artificial Intelligence Generated Content tools, such as AI-assisted programming tools, through semi-structured interviews and interactions with LLMs like ChatGPT-4, Claude, and Gemini. Text analysis was conducted on the collected data. LLMs served as exploratory tools, supplementing in-person interviews and addressing participant scale limitations in traditional qualitative data collection by simulating different professional perspectives.The results indicate that AI tools can enhance development efficiency but require users to have basic programming knowledge for effective interaction. Additionally, designers learning front-end programming can improve cross-disciplinary collaboration and competitiveness, promoting better communication with engineers. This study aims to explore the potential of AIGC tools in assisting UI designers and front-end engineers in transitioning towards the UI engineering field, thereby enhancing their career flexibility and adaptability.

Contribution: This research provides insights into the applications of virtual reality (VR) in learning spatial reasoning, which could be utilized and developed in educational frameworks and settings, especially in science, technology, engineering, arts, and mathematics (STEAM), and other aspects.Background: Spatial reasoning and VR are essential for an effective STEAM strategy. Thus, professionals must constantly research to help learners explore spatial-reasoning perception. This research’s objective is to explore how VR can enhance spatial-reasoning consent of learning, specifically toward mental rotation and spatial visualization skills.Research Question: How does VR help learners embrace spatial-reasoning perception and experiences?Methodology: This study proposes a two-phase comparison experiment to explore the potential improvement of learning spatial-reasoning perception toward spatial abilities and experience in VR. In the first phase, participants experienced conventional hand drawing techniques before moving to the VR environment with the “Gravity Sketch” application, followed by a perceptional assessment in motivated strategies for learning questionnaire (MSLQ) and immersive evaluation with immersive tendency questionnaire (ITQ). In the second phase, participants are invited to conduct the conventional drawing session again, followed by interviews to gain further insights. A galvanic skin response (GSR) device is attached to collect reflection patterns during the whole experiment.Findings: The results support the hypotheses and reveal that VR can help to improve the learning experience and perception of spatial reasoning. Nevertheless, some limitations have been found, such as the small sample size of participants and the need to consider the level of complexity as a future concern.

This study presents an integrated approach to evaluating user experience (UX) in cultural curation by combining big data and thick data analyses, with a focus on the 2023 Hakka Expo in Taiwan (2023 HEIT). Big data plays a pivotal role in identifying UX hotspots and directing future research, while thick data, derived from sources such as user interviews, curatorial proposals, news reports, and social media exchanges, captures the depth of the UX. The fusion of these two data types provides a holistic view of UX, with big data offering quantitative metrics and thick data delivering qualitative insights into emotions and subtleties.Enhanced by a Large Language Model (LLM), the findings reveal that users valued the pragmatic and hedonic design elements of the 2023 HEIT. The LLM's analysis of user feedback and comments provided a detailed understanding of UX, identifying areas of satisfaction and potential enhancements, which aligns with research highlighting LLMs' effectiveness in uncovering user preferences and emotional responses. However, a comparison of high-frequency keywords from thick data sources with those from interview scripts revealed a discrepancy, indicating that the Expo's main messages did not always resonate with visitors. Future research will aim to incorporate more data and a broader range of sources to better understand the combination of big and thick data in studying cultural event experiences.

The rapid integration of Artificial Intelligence (AI) in educational systems has revolutionized teaching and learning methodologies, mainly through the advancement of Generative AI (GAI). This study evaluates the efficacy of emotional blackmail prompts-a novel interaction strategy designed to enhance the responsiveness of large language models (LLMs) like GPT4o, Kimi, and Gemini in educational applications. By leveraging a methodological framework that combines bibliometric and text analysis, our research reveals significant variations in how these models process and respond to emotionally charged prompts. The findings suggest that emotional blackmail can influence the quality and accuracy of AI-generated educational content, highlighting GPT4o's superior ability to adapt to emotional cues compared to other models. This study sheds light on the potential of emotional blackmail prompts to refine AI interactions. It also discusses such strategies' ethical implications and practical applications in improving AI-driven educational tools.

Collaborative brainstorming harbours various positive effects: Enhancement of creativity and social skills, broader discussions and contributions, and instant feedback [ 28 , 46 ], while the technique of mind mapping simultaneously visualises the results of this process. Using a Virtual Reality (VR) application, a technology increasingly adopted for ideation [ 23 , 44 ], this study creates a setting that allows collaborators to produce meaningful results in an immersive digital environment. By nature, group settings remain complex and dynamic, with emotions playing a significant role in the outcome [ 2 ]. So far, emotional responses have mainly been researched through biophysical responses on a single-user basis [ 13 ]. To assess the complex relationship between emotional intelligence (EI) and the language-based results of collaborations, a mind-mapping task was analysed through performance and sentiment analysis, a natural language processing (NLP) technique that identifies the polarity of a given text. We examine the outcome of 13 sessions (N=39) in VR by distinguishing the results into problem-orientation and solution-orientation before applying a fine-tuned language model to get detailed information on the emotional polarity of the results. This mixed-level data analysis shall bridge the gap between self-assessment questionnaires and support the automated group work evaluation by analysing results on an objective scale. Although enhanced problem orientation could not be connected to specific emotions in the sentiment analysis, our results have shown significant relationships between the number of solutions created and the emotions of joy and surprise and a significant negative relationship with the emotion of sadness.

Spatial reasoning is the ability to process and understand spatial information and is important in many fields, including science, art, engineering, architecture, games, etc. This ability can be improved through learning and practice and plays an important role in problem-solving and spatial thinking. The main purpose of this research is to investigate whether there could be a significant improvement in spatial reasoning learning through manipulation exercises in immersive Virtual Reality (VR) environments. We divided our experiment into three phases. The first part was a traditional paper test. Then, followed by a VR environment for immersive drawing practice, and took an immersion level questionnaire (ITQ) after completion. Finally, we returned to the traditional paper test to compare and evaluate the learning progress. During the experiment, the GSR sensor (Galvanic Skin Response) was worn throughout the experiment to collect the participants’ skin electrical responses to correlate with the emotional status of the participants with different educational backgrounds during the experiment. The pilot results of this research show the immersive VR practice can improve spatial learning capability. But there is no significant difference of the participants with different educational backgrounds in the improvement.

Social media design is directly affected by social changes, media development, and advances in hardware devices, and there is a long-term relationship between devices and human-computer interaction design. With widespread use and technological evolution, it has become an interesting research issue and has important substantive applicability to research. Investing in this research topic can produce diverse and very practical research results, and help improve the current inefficient digital advertising pricing method and improve the design method of social media advertising, so as not to be interfered by negative design product. The research results provide positive design and positive use experience. The research outcomes should be able to solve the long-term problem that the delivery benefits of social media advertising pricing in the electronic market are not equal to the actual advertising benefits.