The increasing use of artificial intelligence chatbots in tourism is reshaping how tourists process information and make destination decisions. This study adopts a communicative-goal perspective to conceptualize chatbot language style as either informative or persuasive and examines their differential effects on tourists’ visit intention. Drawing on language expectancy theory and evidence from three experimental studies, the findings show that persuasive language is more effective than informative language in enhancing visit intentions. Further analyses reveal that emotional arousal serves as a key mediating mechanism in the relationship between language style and visit intention. In addition, the findings identify important boundary conditions: tourists with lower levels of AI literacy are more responsive to persuasive language, whereas those with higher perceived risk exhibit a stronger preference for informative language. By extending language expectancy theory to AI-driven communication contexts, this study deepens understanding of chatbot–tourist interactions and highlights language style as a critical design element in AI-assisted destination marketing.
As one of the core businesses for operators, the Point-of-interest Recommendation Problem has attracted widespread attention. Current research is mostly user-centric and confined to deterministic scenarios, whereas, in reality, the operator’s strategic decision-making process faces high demand uncertainty that makes accurate data acquisition extremely challenging. Firstly, we propose a novel distributionally robust model for the Operator-oriented Point-of-interest Recommendation Problem (OO-POIRP) with demand uncertainty, aiming to assist operators in minimizing cooperation cost, ticket booking cost, promotion cost, and transportation cost. Subsequently, the model is reformulated into a tractable mixed integer linear programming model using duality theory. Numerical experiments illustrate that the proposed model outperforms both the stochastic programming model and the deterministic model in terms of risk level by out-of-sample test. In particularly, considering uncertainty can make the model more accurate and credible. Moreover, this paper conducts a case study on the OO-POIRP in Chengdu, China, validating the effectiveness of the proposed model and deriving a series of practically valuable management insights for operators.
Technological revolutions are accelerating the upgrading of tourism technologies. However, existing studies have largely focused on the impacts of external technology (adoption) on tourism, while technological innovation within the tourism industry has long been underestimated. Based on tourism patent data and drawing on evolutionary economic geography, this study examines the evolution of tourism technological innovation in China from 1985 to 2024 and investigates its drivers using province–technology-window panel data for 2005–2024. Descriptive analyses for 1985–2024 show that the knowledge space of tourism technologies has undergone processes of densification and clustering, and knowledge sources exhibit a pronounced trend toward cross-domain integration, with digital technologies playing a prominent enabling role. Tourism technologies remain concentrated mainly in the low-to-medium complexity range and exhibit pronounced regional disparities. Regression analyses for 2005–2024 indicate that regional tourism technology upgrading is path dependent. Market demand reinforces related diversification in tourism technology upgrading, whereas knowledge spillovers attenuate this tendency. This study pioneers a new research field on tourism technology evolution and provides valuable insights into the digital transformation of tourism.
The digital revolution has transformed travel behaviour, with social media and influencer-generated content emerging as dominant forces shaping destination choices. In Kalimpong, homestay tourism has witnessed rapid growth driven by digital persuasion, particularly among young tourists. This study investigates how digital influencers and online visual content affect tourists’ attitudes, satisfaction, and decision-making processes. It examines key constructs such as influencer credibility, visual quality, destination awareness, tourist evaluation, and booking intention. A structured questionnaire based on validated measurement scales was administered to 480 tourists, and data from 465 valid responses were analysed using PLS-SEM. The results demonstrate strong reliability and reveal significant relationships between influencer-generated content, tourists’ perceptions, and behavioural outcomes. Visual richness, perceived credibility, and social media engagement notably enhanced destination awareness and booking intentions. Age emerged as a moderating factor: tourists aged 20–29 were highly active on Instagram and YouTube, driven by visual narratives, fear of missing out (FOMO), and influencer authenticity, whereas tourists aged 30–44 relied more on Facebook and TripAdvisor due to their emphasis on safety, reviews, and structured planning. The seasonal sampling strategy ensured robust temporal representation and minimized bias in travel behaviour. The validated structural model confirms that digital persuasion—particularly credible influencers supported by compelling visual content—substantially shape tourist attitudes and visitation behaviours in the homestay sector of Kalimpong. This multidimensional analysis contributes to a deeper understanding of persuasion dynamics in digital tourism, integrating attitudinal, behavioural, and contextual determinants. Theoretically, this study advances digital persuasion research by extending influencer credibility and visual persuasion frameworks to the underexplored context of community-based homestay tourism and by examining the moderating roles of age and network accessibility. Practically, the findings provide actionable insights for destination managers, homestay associations, and policymakers seeking to design targeted influencer marketing strategies, strengthen digital inclusion, and enhance destination competitiveness in emerging tourism regions.
Hotel booking cancellations pose a persistent challenge for revenue management, leading to demand uncertainty and revenue loss. This study proposes a two-stage deep learning framework that combines TabNet classification with DeepHit survival analysis to jointly estimate cancellation probability and time-to-cancellation risk. The framework is evaluated on the Hotel Booking Demand dataset, comprising 119,390 reservation records from a resort hotel (H1) and a city hotel (H2). In Stage 1, the proposed TabNet classifier achieves the best ranking performance on both datasets, with AUCROC of 0.862 and AUCPR of 0.749 on H1, and AUCROC of 0.846 and AUCPR of 0.827 on H2, outperforming logistic regression, SVM, random forest, gradient boosting, XGBoost, LSTM, and CNN baselines on these metrics. In Stage 2, the DeepHit survival model achieves a concordance index of 0.812 and an integrated Brier score of 0.079, surpassing Cox proportional hazards, random survival forest, and DeepSurv baselines. Beyond aggregate metrics, the framework provides horizon-specific cancellation risk probabilities at 7-, 14-, and 30-day planning windows to support tiered revenue management interventions including dynamic repricing, targeted retention offers, and overbooking adjustment. The results demonstrate that integrating probability estimation with temporal risk modeling provides more actionable decision support for hospitality revenue management than binary classification alone.
With the rapid advancement of digital technologies, hybrid working practice has become the new normal in online travel agencies. However, limited research has explored Gen Z employees’ perceptions and coping strategies regarding this transition. Through an exploratory qualitative research based on 35 in-depth interviews and network text data, this study reveals the intricate dynamics of hybrid working arrangements that integrate both on-site and remote elements. These arrangements significantly reshape Gen Z employees’ perception of work paradigm changes, either catalyzing driving perceptions or triggering inhibiting outcomes. In response, they deploy three distinct coping strategies: problem-focused, emotion-focused, and maladaptive. The study advances an integrative conceptual framework that elucidates the dynamic and reciprocal relationship between Gen Z employees and the evolving organizational work paradigms. Furthermore, the evidence-based recommendations empower organizations to enhance digital technology management and human resource practices, thereby bolstering workforce sustainability and resilience.
This study explores the emotional impact of two distinct types of social media travel videos, induced (professionally produced) and organic (user-generated), on viewers’ destination preferences. Emotional engagement was measured using Emotiv EEG technology, capturing valence and arousal responses from 30 participants aged 18 to 26 as they viewed tourism videos featuring Rome and Barcelona. A mixed experimental design was used. Findings revealed a significantly stronger emotional preference for induced content, suggesting that professionally curated narratives may be more effective in shaping travel intent. Concordance between emotional engagement and destination choice was 70
The resurgence of group tourism highlights a growing demand for meaningful social connections, yet traditional administrative grouping often results in interpersonal misalignment. Drawing on Task-Technology Fit Theory and Social Identity Theory, this research investigates how AI-enabled travel companion matching facilitates “digital co-presence” to enhance relational outcomes. Across three experimental studies, we demonstrate that AI matching significantly outperforms conventional methods in bolstering anticipated social interaction satisfaction. This effect is driven by two parallel socio-psychological pathways: the cognitive mechanism of group identification and the affective mechanism of tourist rapport. Crucially, moderated mediation analysis reveals that the efficacy of these pathways is context-sensitive: cultural destinations—characterized by cognitive goals—amplify the mediating role of group identification, whereas the influence of tourist rapport remains a context-invariant foundational requirement. These findings transition the tourism literature toward a human-human compatibility lens, offering actionable insights for designing preference-aligned algorithms that enhance experiential equity in the algorithmic age.
The increasing availability of user-generated content on review platforms and tourism websites has made sentiment analysis a vital tool for understanding customer experiences. In the hospitality and tourism sector, reviews often contain diverse opinions about specific aspects of service, such as room quality, staff behavior, and amenities. Aspect-Based Sentiment Analysis (ABSA) addresses this by identifying sentiment polarity related to distinct aspects within a review. However, challenges such as implicit aspect references, opinion ambiguity, and context-dependence hinder the effectiveness of existing models. To overcome these limitations, this study proposes a novel ABSA framework based on SpanBERT–Cross-Attention–BiGRU. SpanBERT is employed to extract aspect and opinion terms at the span level, capturing richer contextual dependencies than token-level models. A cross-attention mechanism is then applied to model fine-grained interactions between aspect and opinion spans. The fused representations are passed to a Bidirectional Gated Recurrent Unit (BiGRU) to capture sequential dependencies and predict sentiment polarity. The proposed framework is evaluated on four benchmark tourism and hospitality review datasets, achieving an average accuracy of 94.21
Generative artificial intelligence (AI) tools such as ChatGPT are increasingly used for tourism planning, but they can also produce false or misleading information. This study uses a risk-benefit account to examine how prior AI usage, AI familiarity, perceived ease of use, perceived usefulness, and attitude (the benefit side) combine with AI misinformation/error concern (the risk side) to shape behavioral intention to use AI for tourism planning. Based on public survey data from 900 consumers, the study estimates a confirmatory factor analysis (CFA) measurement model and a structural equation model (SEM) with robust maximum likelihood and full-information maximum likelihood (FIML). The main model, grounded in the Technology Acceptance Model (TAM) and Theory of Planned Behavior (TPB), shows a strong benefit pathway: prior usage predicts familiarity, familiarity predicts ease of use, ease of use predicts usefulness and attitude, and both usefulness and attitude predict intention. Familiarity is also positively associated with AI misinformation/error concern, indicating that more familiar users are more aware of possible AI errors. However, that concern has no significant direct association with intention, and robustness checks do not support any moderation by concern. The study contributes a risk-benefit account in which AI error awareness coexists with positive adoption beliefs but does not outweigh them in this lower-stakes, verifiable tourism-planning context.
This study presents a systematic literature review at the intersection of social media and environmentally sustainable tourism from the tourist perspective. Drawing on a rigorous content analysis of 163 publications, we systematize and synthesize existing knowledge, identify key research gaps, and outline directions for future research. We demonstrate how user-generated social media data are employed to analyze and manage tourist flows at environmentally sustainable destinations. We further examine how contemporary research investigates the influence of social media on destination image and travelers’ behavioral intentions. In addition, we explore how social media are used to examine tourists’ perceptions, experiences, and values related to nature-based and sustainable destinations. Finally, we analyze how social media are utilized to shape tourists’ environmental mindsets. Building on these insights, we advance an integrated conceptual model illustrating how social media function as a socio-digital infrastructure shaping environmentally sustainable tourism. Our findings provide theoretical, methodological, and practical contributions while also demonstrating how social media can be leveraged by scholars and practitioners to promote environmentally sustainable tourism.
Immersive technologies have rapidly evolved and become increasingly embedded in cultural heritage tourism (CHT), providing innovative means to enhance visitor engagement, interpretive experiences, and heritage learning. This systematic literature review critically analyzed 54 peer-reviewed empirical studies published between 2015 and 2025 to explore the shifting dynamics of virtual immersive experiences (VIEs) within the CHT context. This study follows PRISMA guidelines for identification, screening, and inclusion of relevant articles, while thematic clusters and temporal developments are visualized using VOSviewer. This study examines multiple dimensions for CHT research field, including technological applications, visitor experience constructs, research methodologies, and geographic distributions. The findings of this study indicate a clear shift from technology-centered approaches toward experience-driven design, with core experiential constructs such as immersion, behavioral intention, user experience, satisfaction, and authenticity receiving substantial scholarly attention. However, existing studies remain constrained by a reliance on English-language publications, which may limit the cross-cultural generalizability. Moreover, the dominance of cross-sectional and self-report methodologies restricts understanding of the long-term and dynamic impacts of immersive experiences on visitor behavior. This study highlights critical directions for future research and contributes meaningfully to both theoretical development and practical implementation in the domain of immersive cultural heritage tourism.
Tourism 5.0 demands scalable, secure, and transparent systems to enhance service quality, destination image, and visitor loyalty. Traditional approaches face challenges in integrating Artificial Intelligence (AI) with blockchain due to scalability and computational costs. This paper introduces SQDeChain (SQcoin), a hybrid framework combining lightweight AI models with a sharded, Byzantine Fault-Tolerant (BFT) blockchain. Using DistilDeBERTa for text analysis, MobileViT for image processing, and GRU for behavioural predictions, the framework generates key performance indicators (KPIs), Service Quality Trust Index (SQTI), Destination Image Reliability Score (DIRS), and Tokenised Loyalty Propensity (TLP). These KPIs are verified using a novel ClusterPioneer BFT consensus mechanism and energy-efficient operation. The system was evaluated using real-world tourism datasets from Karachi, showing a 41
Accessibility and inclusion remain persistent challenges in cultural and heritage tourism, particularly as artificial intelligence (AI) increasingly mediates visitor experience and destination management. While prior research has examined AI-enabled personalisation and smart tourism development, less attention has been paid to how accessibility is shaped through the collaborative relationship between human users and AI systems in public cultural environments. This paper presents a design-led review of 66 peer-reviewed journal articles published between 2022 and 2026, identified through a PRISMA-guided search of Web of Science, Scopus, and the ACM Digital Library. Drawing on perspectives from human–AI collaboration, inclusive design, and human–data interaction, the review synthesises how AI-enabled systems interpret user data, mediate cultural experiences, and distribute agency across tourism service ecosystems. Building on this synthesis, the paper proposes the Inclusive Human–AI Mediation (IHAM) model, a theoretical framework that conceptualises accessibility as an emergent outcome of data interpretation, AI mediation, human agency, and governance. The model advances tourism technology research by reframing accessibility as a systemic and design-mediated challenge, and provides guidance for designing inclusive, ethical, and sustainable AI-enabled cultural and heritage tourism systems.
The rapid integration of AI-driven robots into hospitality and tourism services raises critical questions about how communication styles shape consumer behavior. This study investigates how conversational styles (emotional vs. rational) influence consumers’ purchase intentions, examining the mediating role of engagement and the moderating role of robot persona realism. Grounded in the Social Presence Theory and Stimulus–Organism–Response framework, evidence from one field study and three online experiments reveals that emotional styles enhance engagement and purchase intentions, especially when paired with more persona realism. Conversely, rational styles are more effective for less persona realism. These findings advance theories of by demonstrating the importance of aligning communication styles with robot design. For practitioners, the study offers actionable guidance on deploying conversational AI agents that foster engagement, comfort, and conversion, ultimately enriching hospitality and tourism experiences while supporting responsible AI adoption.
Artificial intelligence (AI)–powered recommendation systems are increasingly shaping cultural tourism experiences by personalizing destination suggestions and enhancing informational richness. While AI offers practical benefits such as efficiency and convenience, cultural tourists often experience ambivalence driven by concerns over authenticity loss, privacy risks and algorithmic manipulation. This study investigates the perceptual heterogeneity of outbound cultural tourists toward AI-enabled cultural recommendations and identifies distinct psychological profiles that inform differentiated engagement behaviors. Using survey data from 358 respondents, k-means clustering based on perceived benefits (PB), perceived concerns (PC), engagement intention (ENG), trust and AI familiarity revealed three meaningful segments: AI-Enhanced Cultural Explorers, Selective Cultural Learners and Authenticity-Driven Skeptics. MANOVA results confirmed significant multivariate differences among clusters, while machine-learning validation using Random Forest, SVM, XGBoost and SHAP analysis demonstrated that emotional–cognitive variables—particularly trust and perceived concerns—are the strongest predictors of segment membership. These findings indicate that perceptions of AI in cultural tourism extend beyond functional utility and are strongly shaped by emotional interpretation and authenticity-related values. The results highlight the importance of targeted communication, transparency and culturally sensitive AI design strategies tailored to distinct tourist orientations. The study contributes to theory by advancing perception-based segmentation within AI tourism research and suggests practical pathways for responsible, human-centered AI implementation that supports cultural sustainability.
This paper explores how Flickr’s geotagged photos have contributed to the development of new research topics in tourism studies, particularly through the use of large-scale, freely available datasets. A systematic literature review and bibliometric network analysis were conducted using 333 Scopus-indexed papers that included “tourism” and “Flickr” in abstracts or keywords. An additional 519 papers citing this core set were identified. Network analysis using Gephi applied a gravity model and clustering algorithms to detect citation-based communities. Content analysis of highly cited papers helped define key research themes. Seven research clusters emerged, focusing on nature-based tourism, tourist activities by space-time behaviour, destination attractiveness, image development, travel route detection and recommendation, and machine learning for content analysis. These communities reflect the global academic interest enabled by Flickr’s open API, allowing reproducible and comparative analyses across destinations. This study highlights how a unified, open-access social media dataset has catalysed the formation of global research communities in tourism. Unlike newer platforms with restricted access, Flickr’s openness fostered methodological innovation and deepened field-specific knowledge in tourism research. A critical analysis of existing work was provided, highlighting overlooked areas and synthesising insights to propose new directions for research.
Digital tourism marketing increasingly relies on professional user-generated content (PUGC), yet mechanisms through which multimodal videos and real-time social interactions shape travel decisions remain largely opaque. This study identifies these mechanisms by analyzing 2,650 tourism videos from Bilibili and their synchronized Danmaku (bullet-screen comments) through a computational framework integrating computer vision, natural language processing, and machine learning. We develop and validate five cognitive load metrics through comprehensive construct validity assessment (structural, known-groups, criterion, and nomological validity). Three findings emerge: First, the cognitive interaction term (SD × CD) ranks as the single most important predictor among 95 multimodal features (SHAP importance = 16.2
This study examines how Reddit communities framed the Nipah virus (NiV) outbreak during December 2025–January 2026, with particular attention to “next pandemic?” narratives, airport screening discourse, and misinformation. Analysis of 38 posts and 1,125 comments from news, travel, health/science, conspiracy, and general/country publics reveals distinct ecology-dependent framing patterns. While outbreak geography and transmission routes dominated overall discourse, “next pandemic?” framing was more prevalent in general/country and health/science communities than in travel forums. Airport screening appeared prominently in headlines but less in deliberation, functioning as an ambiguous tourism signal. Misinformation clustered primarily in conspiracy spaces with limited cross-frame coupling. The findings advance tourism crisis scholarship by conceptualizing Reddit as interacting publics that differentially process outbreak narratives, offering implications for travel-risk communication during short-lived attention spikes.
Wellness tourism is a major segment of the global tourism economy, yet many flow datasets lack the theme semantics and behavioral context needed to resolve high-resolution, theme-heterogeneous networks, a limitation that LLM-based parsing of narrative travelogues can help address. We develop and validate a framework integrating large language models (LLMs), social network analysis (SNA), and quadratic assignment procedure (QAP) inference. Using 10,457 travelogues from Yunnan, China (2010–2025), the LLM jointly reconstructs county-level origin–destination (O–D) wellness tourism flows (WTF) and assigns wellness themes. The resulting networks are sparse yet hierarchically organized, with systematic differences across themes. QAP analyses indicate distinct drivers for WTF tie activation versus conditional intensity: activation is constrained by distance, reception capacity, and environmental quality, whereas intensity is more sensitive to theme-specific supply elasticity and market conditions. This study provides a reproducible, text-based workflow for reconstructing theme-labeled wellness tourism flows and supports tourism governance evaluations that account for thematic and stage-specific differences.
