Abstract

Abstract
The rapid evolution of artificial intelligence (AI) has transformed customer service environments, with chatbots emerging as key tools for improving efficiency, responsiveness and personalization. Despite their growing adoption across sectors such as retail, finance, healthcare and education, uncertainty persists regarding how users perceive these systems and which factors shape their satisfaction. Existing research points to clear advantages of chatbots but also highlights limitations related to trust, empathy and the handling of complex interactions.? This paper examines how AI-based chatbots influence user satisfaction in customer service contexts and identifies the interactional factors that most strongly shape users’ evaluations. A quantitative, cross-sectional survey was conducted using a questionnaire that explored perceived interaction quality, trust in AI systems, social presence, usefulness, responsiveness and limitations experienced during automated service. Statistical analysis was used to identify relationships between chatbot use, user satisfaction and the variables contributing most strongly to positive or negative evaluations.? The results show that although most respondents had previous experience with chatbots, their assessments varied across performance dimensions. Ease of use received the most favourable ratings, while problem resolution showed the greatest variability, suggesting that chatbots remain more effective for simple requests than for complex issues. Trust and clarity of information were generally positive but displayed substantial dispersion, reflecting heterogeneous user experiences. More than half of users required human assistance after chatbot interaction, and barriers such as lack of need, privacy concerns and perceived inconsistencies were frequently reported. Despite these challenges, overall satisfaction and recommendation levels were positive, indicating that effectiveness in resolving issues is central to user acceptance.

Abstract

Abstract
The digital transition has transformed the organisational and creative dynamics of communication agencies, generating new market demands and redefining the role of creative professionals. This study investigates how agency creatives perceive the impact of digital platforms and the internet on professional practices, creative autonomy and strategic decision-making. Adopting a qualitative approach, semi-structured interviews were conducted with seven experienced professionals working in Portuguese communication agencies. The corpus was analysed using the Iramuteq software, which enabled Descending Hierarchical Classification, Correspondence Factor Analysis, similitude mapping and word cloud visualisation. Results reveal that digitalisation is not perceived merely as a technological update, but as a structural force that intensifies production cycles, imposes data-driven evaluation and reshapes the symbolic value of creativity. Although digital tools expand expressive possibilities, they simultaneously restrict time for conceptual exploration and increase performance pressures. The findings demonstrate that contemporary creativity has become a negotiation between artistic intention and measurable market outcomes. The study concludes that sustaining creative autonomy requires organisational reform, interdisciplinary learning and strategic anticipation of digital trends. Insights highlight the need for agency models that protect creativity while embracing technological innovation.

Abstract
Context: Mutation testing is an effective fault-based testing criterion increasingly adopted in industrial settings. However, the traditional mutation score does not communicate the effort required to improve test adequacy, creating a governance gap for managers planning and prioritizing testing work. Objectives: This paper formalizes Mutation Technical Debt (MTD) as a conceptual framework that positions mutation testing within technical debt theory, and proposes estimation formulas for its principal—the residual effort required to achieve mutation adequacy. Methods: We define two principal estimation formulas: MTD_manual, capturing human effort in manual testing workflows, and MTD_auto, modeling costs in automated pipelines. A hybrid formula combines both based on an organization’s activity mix. We demonstrate the metric using an empirical study with 39 C programs featuring pre-determined equivalent mutants and adequate test sets, applying incremental test-case addition and non-parametric statistical analysis (Spearman correlation, Friedman test, Wilcoxon signed-rank test). Results: MTD shows a strong negative correlation with mutation score (? = -0.82, p < 0.001), decreases significantly with each incremental test case (Friedman ?² = 113.95, p < 0.001, d = 0.80), and reveals an approximately 26-fold cost difference between manual and automated approaches (p < 0.001, r = 1.0). An SSDL-adequate test set reduces overall MTD across all 74 operators by 86%. ?Conclusion: MTD bridges the gap between testing adequacy and project governance by translating mutation scores into actionable effort estimates, enabling managers to prioritize investment, integrate quality gates into CI/CD pipelines, and support business cases for test automation.

Abstract
Renewable Energy Communities (RECs) are emerging as key enablers of decentralized, sustainable, and consumer-driven energy systems. Beyond environmental benefits, RECs possess significant potential to enhance resilience against extreme weather, price volatility, and infrastructure fragility. This article integrates resilience and relia bility constraints directly into the planning and operation of RECs, assessing their impact on system cost, sizing, and dispatch. Two optimization models are developed: a design model that sizes community assets (PV and BESS) using varying resilience indicators, and an operational model that minimizes costs while monitoring reliability. The analysis introduces two resilience metrics, deterministic hourly autonomy and average autonomy, and eval uates them using real-world data from the Caxias Living Lab. Results demonstrate that average resilience can be increased with minimal cost impacts due to non-linear trade-offs, whereas strict hourly resilience requires signifi cant storage investment. Furthermore, a Value of Lost Load (VoLL) reliability indicator is shown to cost-effectively trigger maintenance events. This framework offers actionable guidance for designing sustainable, adaptive, and economically viable energy communities.