Takming University of Science and Technology

Against the backdrop of rapid development in generative artificial intelligence technology, the definition, process, and evaluation standards of artistic creation are undergoing a profound transformation. This study takes “The Era of Prompts—A Challenge Letter from AI to Humanity” exhibition at Tainan Art Museum as the research field, employing mixed research methods to explore the multiple impacts and educational implications triggered by AI intervention in artistic creation. The research integrates exhibition ethnography, procedural scaffolding experimental design, and computational grounded theory analysis, conducting a three-stage learning journey experiment with 206 university students in southern Taiwan. The primary objective was to understand how university students perceive and engage with AI in the context of art creation, and to identify the core themes that emerge from their learning experiences. The main findings reveal five grounded themes of AI art learning: computational writing practice, emotional expression mediation, imperfection value reconstruction, collaborative relationship dynamics, and literacy requirement identification. These five themes interact spirally to form the theoretical model of “Adaptive Development of Art Learning in the AI Era.” The research concludes that art education in the AI era needs to construct new pedagogical paradigms that embrace technological innovation while maintaining human subjectivity, criticality, and emotional depth in creation. This study provides important references for theoretical construction and educational practice of AI art, offering a nuanced understanding of the human-AI creative partnership. The findings suggest that rather than viewing AI as a mere tool, it should be approached as a collaborator, a mediator of emotional expression, and a catalyst for re-evaluating the very nature of creativity. This research contributes to the fields of art education, human-computer interaction, and digital humanities by providing an empirically grounded framework for designing and evaluating AI-integrated learning environments.

Purpose This study aims to investigate the factors influencing Indonesian consumers’ telemedicine retention intention in the post-COVID-19 period. Whereas most work examines initial adoption, this research emphasizes the drivers of long-term engagement and retention within an emerging market context. Design/methodology/approach Drawing on the Push-Pull-Mooring (PPM) framework, this study conceptualizes telemedicine retention intention as shaped by three sets of factors: push (limitations of traditional healthcare systems), pull (benefits of telemedicine) and mooring (behavioral inertia). Data were collected from 509 Indonesian telemedicine users and analyzed via partial least squares structural equation modeling (PLS-SEM) to examine both direct and moderating effects. Findings The results show that both push and pull factors significantly influence consumers’ intention to continue using telemedicine services. Inertia, defined as attachment to traditional in-person care, exhibits a dual role and in our sample shows a small positive direct association with retention. Moderation analysis further reveals that inertia reduces the influence of push factors while enhancing the effect of pull factors. These findings highlight user convenience, affordability and technological reliability as crucial for maintaining long-term engagement with digital health services. Originality/value This study extends the application of the PPM model to the healthcare domain in an emerging market context, offering insights into postadoption behavior in digital health. It clarifies the dual role of consumer inertia and proposes a habit-driven perspective on telemedicine retention. The findings advance theory and inform practice for sustaining digital healthcare services in Southeast Asia.

This study explores the development of intelligent demand forecasting methods in the context of the rise of rational consumption awareness, to establish the optimal spare parts reuse plan. As consumers pay more attention to sustainability and environmental protection, price and quality are no longer the only considerations; global inflation has also made consumer behavior more conservative, emphasizing budget control and long-term value. In response to the sluggish market demand, this study proposes a intelligent demand forecasting method to help companies understand consumer demand and design product reuse plans that are in line with their values. This study adopts the concept of "integrated forecasting" and designs two methods: (1) Hybrid Stacking (HS) Method: Combining traditional time series and machine learning techniques to improve forecast accuracy; (2) External Information Integration (EI-HS) Method: Incorporating external information into the HS model as an improvement baseline to further reduce forecast errors. Finally, the results of MASE and RGRMSE indicate that the methods proposed in this study hold strong application potential. The research results emphasize that the development of intelligent demand forecasting methods in the era of rational consumption is of key significance and points out the direction of sustainable improvement and expansion in the future, providing companies with more accurate forecasting tools and spare parts reuse decision-making basis.

Internet of Things (IoT) technology has become an integral part of human life, aiming to facilitate and increase the efficiency of human operations. The intelligent home system, which adopts IoT technology, can improve security, convenience, and efficiency. In this study, the intelligent home system will be developed, which employs several physical sensing devices, an ESP32 microcontroller, ThingSpeak, and Line as notification control, and MQTT as the communication protocol. System development refers to the IoT architecture, which consists of three layers, including the physical layer, network layer, and application layer. System testing is carried out by capturing the real-time condition parameters of humidity, temperature, and smoke concentration through the ThingSpeak app, which indicates that all of the parameters can be read by the IoT system. In addition, Line Notify apps enable sending real-time message push notifications to the user regarding the status updates of smoke concentration, door lock status, and live video streaming. In conclusion, the intelligent home system developed in this study provides the enhancement of home security system, user convenience, efficient device, and low-cost system.

Abstract With the rapid adoption of “Internet Plus” technologies in the tourism industry, online purchasing of travel products has become increasingly common, and user-generated reviews now play a vital role in consumer decision-making. This study develops a customer satisfaction evaluation model for customized online tour packages by combining text mining, Latent Dirichlet Allocation (LDA), and focus group interviews. The DEMATEL and DANP methods are used to examine the relationships and relative importance among key factors. Results show that product experience is the most influential factor affecting customer satisfaction, with product attributes and service quality having a significant impact on the overall experience. Based on these findings, the study provides targeted recommendations for product improvement. It suggests future research directions, including the integration of data from multiple platforms and the conduct of behavioral validation to enhance the model’s usefulness.

This systematic literature review examines the role of ChatGPT in communication. ChatGPT's ability to imitate human-like interactions has broad implications in various sectors, such as education, healthcare, and customer service in the digital-based economy. The authors used a systematic and structured manuscript selection method in this research to collect and analyze literature on the use of ChatGPT in a communication context. A systematic literature review (SLR) method was used, involving an extensive search through the Scopus and Google Scholar databases with the keywords "ChatGPT" and "communication." Manuscript selection required strict inclusion and exclusion criteria. Of the 623 articles found, 30 were selected for further review. The research results show that using ChatGPT in communication has had both positive and negative impacts. Positive impacts involve increasing the efficiency and effectiveness of communications, especially in education, marketing, ethics, and health. However, challenges such as ethical considerations, the risk of plagiarism, and a limited understanding of context and emotional interactions were also identified. The use of ChatGPT in education, health, and various other fields has demonstrated great potential to improve communication processes, decision-making, and work efficiency. However, to ensure responsible and sustainable use, we must address specific ethical challenges and risks. This study provides a comprehensive overview of recent developments in using ChatGPT in communications, while also highlighting the practical and ethical implications that must be considered. With careful consideration of the advantages and limitations, ChatGPT in communications can significantly contribute to various fields.

In a real business world, the production plan must respond to quick changing market and keep production line stable. An optimal stepwise production strategy is derived with stable production rate for some period, and then changing to another stable production rate to response market change for the next period. The cost factors we considered in the model include the production setup cost, the production-variation cost, the carrying cost, the deteriorating cost, the backordering and the lost sale shortage cost. For demand that changes linearly or exponentially with time, an inventory model with partial backordering is developed. Finally, sensitivity analysis is carried out to investigate how changing parameters affect this system.

This study proposes a Bayesian test for a test portfolio p’s mean-variance efficiency that takes into account the sampling errors associated with the ex-post Sharpe ratio ŜR of the test portfolio p. The test is based on the Bayes factor that compares the joint likelihoods under the null hypothesis H0 and the alternative H1. Using historical monthly return data of ten industrial portfolios and a test portfolio, namely, the CRSP value-weighted index, from January 1941 to December 1973 and January 1980 to December 2012, the power function of the proposed Bayesian test is compared to the conditional multivariate F test by Gibbons et al. (1989) and the Bayesian test by Shanken (1987). In an independent simulation study, the performance of the proposed Bayesian test is also demonstrated.