Given the uneven financial resources, digital capabilities, technical support, and production environments characterizing Thai smallholder agriculture, Internet of Things (IoT) adoption readiness requires both willingness to adopt and confidence in using IoT under actual farming conditions. This study examines the determinants of IoT adoption readiness using an extended UTAUT2-based framework that distinguishes behavioral intention from technological self-efficacy. The framework was extended by incorporating government support and financial support as two context-specific factors, while technological self-efficacy was positioned as a capability mechanism connecting enabling conditions with behavioral intention. Data were collected through face-to-face questionnaires from 475 smallholder farmers across four regions of Thailand and analyzed using covariance-based structural equation modeling. The structural model demonstrated acceptable fit (CFI = 0.934, RMSEA = 0.059, SRMR = 0.050) and explained 54.7% of the variance in behavioral intention and 72.7% in technological self-efficacy. Technological self-efficacy was the strongest predictor of behavioral intention (β = 0.393, p < 0.001), followed by habit and price value. Environmental suitability was the strongest predictor of technological self-efficacy (β = 0.493, p < 0.001), followed by financial support and facilitating conditions. In contrast, effort expectancy did not significantly influence technological self-efficacy. The findings demonstrate that agricultural IoT readiness depends not only on perceived benefits and economic value but also on farmers’ confidence that IoT technologies can be applied within their actual production environments. The study highlights the importance of locally compatible technologies, capability development, financial accessibility, and coordinated institutional support for promoting agricultural IoT adoption among Thai smallholders.
Business intelligence technologies have evolved from retrospective reporting tools into strategic assets integrating artificial intelligence, cloud computing, and big-data infrastructure. However, systematic, evidence-based roadmaps for forecasting BI innovation trajectories remain limited. This study investigates principal technology clusters within BI innovation, their lifecycle stages, and maturation timelines relevant to strategic planning and open innovation decision-making. Using a multi-stage patent analytics framework, 14,072 BI patents retrieved from Lens.org and spanning 1928–2026 were analyzed. Four transformer-based sentence-embedding models were evaluated to capture semantic relationships, and ensemble validation selected architectures suited to the patent corpus. Uniform Manifold Approximation and Projection reduced embedding dimensionality, while K-means clustering was assessed across 4–25 clusters using the Silhouette coefficient, Davies–Bouldin index, Calinski–Harabasz score, and Gap statistic. The analysis identified 20 technology clusters representing infrastructure platforms, analytical methods, and vertical applications. Cluster themes were characterized through keyword extraction combining class-based TF–IDF with BM25 Okapi scoring, revealing technologies across healthcare, telecommunications, retail, financial services, and smart cities. Logistic S-curves fitted to cumulative patent counts estimated lifecycle stages and projected saturation years, with partial-year data treated explicitly to avoid truncation bias. All clusters were classified as emerging or growing based on patent activity trajectories; two are projected to reach saturation by 2033, whereas eighteen extend beyond 2034. The resulting multi-horizon roadmap supports research prioritization, investment allocation, competitive positioning, external technology search, ecosystem mapping, and partner identification for organizations pursuing open innovation strategies. Methodologically, the study advances patent analytics through comparative transformer evaluation, multi-metric cluster validation, and fused keyword extraction.
This study examines how entrepreneurial motivation relates to venture performance among 120 Moroccan entrepreneurs supported by the National Initiative for Human Development (INDH). Respondents were positioned on a necessity (Push)–opportunity (Pull) continuum on the basis of fourteen motivation items, and the resulting profiles were analyzed with statistical tests and an explainable machine-learning pipeline combining eight classifiers evaluated by repeated stratified cross-validation, SHAP attributions and kernel principal component analysis. Pull entrepreneurs were less numerous (38.33%) but more often reported a satisfactory income than Push entrepreneurs (91.30% vs 63.51%; odds ratio = 6.03, p < 0.001) and had a higher composite performance score (Cohen’s d = 0.83). Because the Push/Pull label is derived from the motivation items, the classifiers are used as an interpretive device rather than as predictive tools; all of them reached cross-validated accuracies between 0.92 and 0.97, far above a majority-class baseline (0.62). SHAP attributions show that survival income and family expectations are the motives that most sharply separate the two profiles, while kernel principal component analysis shows that experience, monetary aspirations and relationships with previous employment structure heterogeneity within each profile. Opportunity recognition, earning more money, autonomy and self-realization are positively associated with performance, whereas survival, previous unemployment and family expectations are negatively associated with it. These cross-sectional associations are not causal, but they support a differentiated, motivation-sensitive design of entrepreneurship support programs, to be validated through prospective evaluation.
Open innovation research has expanded beyond firm-centred R&D, yet its operation in low-formality, community-based tourism networks remains underexplored. This study examines how cultural tourism and ecotourism are linked through open innovation in Pattani, southern Thailand, a culturally embedded and security-constrained context. It investigates how entrepreneurial human capital, digital capability, and collaborative capability relate to open innovation capability (OIC), and how OIC is associated with service integration capability (SIC) and community tourism enterprise performance (CTEP). An exploratory sequential mixed-methods design was employed. Phase 1 used 23 in-depth interviews to refine the constructs. Phase 2 analysed survey data from 302 community-based tourism entrepreneurs and local service providers using PLS-SEM with a higher-order specification of OIC. Phase 3 involved structured validation of the findings by nine experts. All three micro-foundations were positively associated with OIC. Collaborative capability had a coefficient of 0.329; pairwise tests indicated that its association with OIC differed significantly from digital capability, but not entrepreneurial human capital. Digital capability showed a lower coefficient than anticipated, possibly reflecting uneven relational and infrastructural conditions. OIC was positively associated with SIC, and both were positively associated with CTEP. The mediation and serial mediation patterns were consistent with the proposed capability-chain mechanism, although the cross-sectional design does not establish temporal or causal mediation. The study extends open innovation theory to low-formality, culturally embedded networks, conceptualizes OIC as a multidimensional network-level construct, and positions SIC as a potential operational link between knowledge flows and perceived performance.
Patent analysis often supports patent-filing decisions, technological forecasting, and the identification of research gaps. Although patent applications are widely used as indicators of innovation output, less attention has been paid to tracing the geographic origin of inventive activities, which can inform forward-looking investments and R&D resource allocation decisions. To address this gap, this study analyzes technological emphasis in internationally filed patent families by comparing Japan-priority and U.S.-priority patent families in the JaParaPat (Japanese-English Parallel Patent Application) corpus. We identify systematic differences in technological emphasis using sentence embeddings, clustering, and BERTopic modeling. The results indicate that the Japan-priority patent families are more strongly represented in engineering-intensive domains, including hardware, control systems, and manufacturing technologies, whereas U.S.-priority patent families are more concentrated in software-driven, data-centric, and biomedical domains. These findings provide empirical evidence of differences in the representation of technological domains across priority-filing countries, specifically among patent families in which protection was sought in both Japan and the United States.
