Traditional villages embody tangible repositories of historical, cultural, and geographical heritage, and their sustainable and authentic development poses a global challenge. By applying complex adaptive system (CAS) theory via a bottom–up approach, we analyze traditional settlements using China’s Zheshui village as a representative case. Road networks and spatial configurations were examined through image analysis (ImageJ 1.54 p, Depthmap+ Beta 1.0), integrating space syntax, box-counting dimension, and point-density analysis to decode hierarchical point-line-plane structures. Key findings reveal that building units self-similarly aggregate into courtyards under landmark constraints, with courtyards further coalescing into villages. Road systems function as adaptive agents that facilitate nodal information flow while exhibiting fluidity and diversity. The village emerges as a macro-scale complex system from the building-unit level, displaying cross-scale self-similarity, yet intrinsic diversity in architecture and roads underlies its core complexity. BTM topic modeling of tourist sentiment—identifying tourists as novel adaptive agents—predictively guides strategies for enhanced cultural dissemination and public infrastructure. By establishing a CAS-driven internal generative mechanism, this work offers a novel methodological framework for authentic conservation and sustainable development.

The pursuit of autonomous chemical transformations with single-bond precision represents a central challenge in molecular nanoscience. While scanning tunneling microscopy (STM) enables site-specific reactions by directly engaging individual atoms and bonds, conventional approaches rely on expert intervention and lack reproducibility and scalability. Here we introduce a deep learning-based strategy that autonomously executes multi-step, bond-selective transformations. Our system integrates computer vision for molecular recognition, neural networks for bond-state classification, and deep reinforcement learning for closed-loop optimization of activation parameters. As a proof of concept, we demonstrate the selective dissociation of C-Br bonds in a tetra-brominated porphyrin on Au(111). Importantly, the approach extends beyond single-bond events, enabling programmed multi-step sequences including four distinct pathways with high fidelity. By advancing from isolated, human-directed manipulations to fully autonomous, data-driven reaction control, this platform establishes a paradigm for intelligent single-molecule chemistry. It provides a generalizable framework for on-surface synthesis, where adaptive agents orchestrate molecular transformations with a level of precision and scalability unattainable by manual approaches.

Calcium carboxymethyl cellulose/quaternary ammonium chitosan self-gelling powder with good biocompatibility for wound hemostasis.

Advancing hierarchical optimization: A-Cubed algorithm for adaptive agent collaboration

Artificial facial mimicry (AFM) is increasingly used to enhance social interaction with virtual agents in immersive virtual reality. However, its psychological and ethical implications remain insufficiently explored. This article conceptualizes AFM as an effective and embodied intervention, examining the role of emotional congruence, individual differences, and clinical vulnerability in shaping user responses. We further outline methodological directions involving physiological measures and embodied coordination. By framing AFM within affective computing and embodied cognition, this work contributes to the responsible design of emotionally adaptive virtual agents.

Intelligent Tutoring Systems are increasingly used in higher education to support personalized learning and academic monitoring in large-scale digital environments. However, existing systems are predominantly based on static architecture and rigid rule-based mechanisms, which limit scalability and hinder effective adaptation to heterogeneous learners, evolving learning behaviors, and real-world educational contexts. This paper presents a self-adaptive multi-agent architecture based on Reinforcement Learning for autonomous decision-making in intelligent systems deployed in real environments. The proposal integrates an RL Meta-Agent that dynamically optimizes the selection of specialized agents through an intelligent switching mechanism, considering the user’s state, behavior, and interaction patterns. The architecture was implemented in Moodle using flows orchestrated in n8n, LLMs, databases, APIs developed in Django, and real academic data. For the empirical evaluation, a real and a simulated case study were designed. A questionnaire was administered to university students, considering dimensions of usability, satisfaction and usefulness, and accessibility and interaction, to understand the perception of the system and improvements. The quantitative data were analyzed using descriptive statistics and nonparametric tests (Mann–Whitney U and Kruskal–Wallis), while the qualitative data were examined using thematic categorization. A simulated case study was conducted to analyze the behavior of the system. The results show that the RL Meta-Agent significantly improves system efficiency, response relevance, and adaptive agent selection, demonstrating that self-adaptive RL-based MAS architectures are a viable solution for intelligent systems applied in real-world contexts, providing empirical evidence of their performance and adaptability in complex scenarios such as higher education.

Orientation: Workplace inclusivity is increasingly prominent, yet Gen Z’s contribution to disability inclusion in small and medium enterprises (SMEs) within developing contexts remains underexplored. Research purpose: This study aimed to examine Gen Z employees’ perceptions of inclusivity, the influence of gender-awareness on fairness, and organisational strategies supporting workers with disabilities. Motivation for the study: Prior work prioritises large corporations and managerial views, overlooking generational differences and gender–disability intersections. Gaps are critical in Indonesia’s structural and cultural context. Research approach/design and method: A qualitative design was used across Pontianak, Palangka Raya, and Bandar Lampung, combining semi-structured interviews, open-ended surveys, and document analysis on workers with disability, peers without disability, mentors, managers, and human resource (HR) officers. Data were thematically analysed (Braun & Clarke). Main findings: Gen Z acted as agents of adaptation, using digital tools, basic sign language, and teamwork to bridge communication and mobility barriers. Fairness was gendered: women often faced over-assistance; men were pushed towards hyper-independence. Organisational efforts (flexible scheduling, awareness training, mentorship) helped, but inaccessible infrastructure (e.g. lack of ramps or lifts) remained a major constraint. Practical/managerial implications: Institutionalise peer-level support, implement gender-sensitive HR practices, and prioritise infrastructural accessibility. Contribution/value-add: This study integrates generational and gender perspectives into disability inclusion, offering actionable guidance for SMEs in resource-constrained settings to move from symbolic compliance to systemic equity.

ABSTRACT Rural students often face limited exposure to multidisciplinary knowledge and insufficient training in complex problem-solving (CPS) skills. To address this gap, this quasi-experimental study adopted a progressively integrated multidisciplinary STEM approach based on the Four-Component Instructional Design (4C/ID) model, with Large Language Models (LLMs) embedded as adaptive scaffolding in one experimental condition. A total of 120 Grade 6 students from rural schools were randomly assigned at the class level to three conditions: AI-enhanced 4C/ID scaffolding (EC1), 4C/ID scaffolding without AI (EC2), and traditional instruction (CC). A standard K-6 Earth systems unit was reorganized into four sequential STEM tasks. Data were collected through CPS performance tasks, conceptual quizzes, and questionnaires on cognitive load and learning satisfaction. One-way ANOVA, post-hoc tests, and mediation analysis were conducted. Results showed that EC1 significantly outperformed EC2 and CC across all outcome measures, while EC2 also exceeded CC, underscoring the value of structured scaffolding. Mediation analysis further indicated that reduced cognitive load partially explained the satisfaction gains in EC1. These findings suggest that LLMs can function as effective adaptive scaffolding agents to support equitable and inquiry-driven STEM learning for under-resourced rural learner.

The digital transformation of communication has significantly impacted religious outreach practices, including those of modern Islamic movements such as Muhammadiyah. This article examines the implementation of a digital-based da’wah training program for Muhammadiyah cadres in Tosari, Pasuruan, East Java. The program aimed to enhance participants’ digital literacy, content creation skills, and contextual communication strategies grounded in Islamic values. Utilizing a participatory and experiential training model, participants engaged in hands-on workshops and mentoring sessions focused on tools such as Canva, CapCut, and social media platforms like Instagram, TikTok, and YouTube. Evaluation results indicated substantial improvements in digital skills, increased confidence in engaging youth audiences, and a paradigm shift among participants regarding the use of technology for religious outreach. The findings suggest that integrating religious education with media literacy can empower da’wah practitioners as adaptive change agents within the digital public sphere. This model holds potential for replication in other regions, with contextual adjustments to meet local needs.

This study examines decision-making in intelligent virtual agents (IVAs) and formalizes the distinction between tactical decisions (individual actions) and strategic decisions (composed of sequences of tactical actions) using a mathematical model based on set theory and the Bellman equation. Although the equation itself is not modified, the analysis reveals that the discount factor (γ) influences the type of decision: low values favor tactical decisions, while high values favor strategic ones. The model was implemented and validated in a proof-of-concept simulated environment, namely the Snake Coin Change Problem (SCCP), using a Deep Q-Network (DQN) architecture, showing significant differences between agents with different decision profiles. These findings suggest that adjusting γ can serve as a useful mechanism to regulate both tactical and strategic decision-making processes in IVAs, thus offering a conceptual basis that could facilitate the design of more intelligent and adaptive agents in domains such as video games, and potentially in robotics and artificial intelligence as future research directions.

This study presents TAQLA, a new Tabular Adaptive Q-Learning Agent for portfolio management in stochastic financial markets. TAQLA rests on a multi-model reinforcement learning (RL) architecture that integrates parameter-adaptive Q-Learning mechanisms into softmax-based exploration to reconcile short-term profit maximization with long-term capital preservation. The method is contrasted with vanilla Q-Learning, SARSA, and a random trading policy using simulated equity market data. Empirical analysis shows that TAQLA performs better on profitability, risk-adjusted performance, and drawdown minimization, with a last portfolio value of $1687.45 (+68.74% of initial capital), a Sharpe ratio of 1.41, and a maximum drawdown of just 12.8%. Q-Learning and SARSA, on the other hand, yield Sharpe ratios below 1.0 and drawdowns exceeding 18%. Parameter sensitivity analysis across β (softmax temperature), α (learning rate), and γ (discount factor) reveals that aggressive exploration (β ≈ 1.0–1.5) and reasonable discounting (γ ≈ 0.4–0.6) generate the most aggressive and robust outcomes. Such outcomes place TAQLA as a robust RL-based adaptive portfolio control method under uncertainty, with improved capital appreciation and robustness to adverse market conditions.

This paper introduces an innovative intelligent system called Adaptive Packing Intelligence (API) for spatial optimization. It builds upon traditional packing models by enabling agent-controlled interactions among geometric objects. Unlike standard methods that enforce strict non-overlap constraints, API offers flexible control over object placement through dynamic agent parameters. These parameters are integrated into phi-functions within the mathematical framework, regulating spatial relationships and supporting three modes: enforced separation, exact contact, and controlled non-overlapping. The system's architecture features a unified mathematical core based on phi-functions for geometric representation, an adaptive Agent for managing real-time interactions, and an optimization engine utilizing gradient-based and heuristic algorithms to identify optimal configurations. API can address various objective functions, including volume minimization, packing density maximization, and energy-based interaction modeling. Its applications extend across medicine, biology, and engineering. By unifying diverse disciplines under a single framework, API enhances adaptability and computational efficiency in solving complex packing and simulation problems

As habitat fragmentation and land-use changes increase alongside increased human activities, human wildlife conflict has risen and is currently posing dire consequences for biodiversity conservation and ecosystem stability. This paper presents an analysis, prediction, and mitigation of human-wildlife conflicts using a multi-agent system (MAS) simulation framework to facilitate ecosystem restoration planning. It is founded on the assumption that human beings, wildlife, and environmental units are represented as adaptive autonomous agents with adaptive behaviors as well as decision-making capabilities. The primary objective is to measure the impacts of agent interactions, movement, and resource competition patterns on the prevalence of conflict and habitat recovery. The spatial land-use, ecological properties of species, and patterns of human activity within an agent-based simulation environment form the methodology for experimenting with scenarios combining diverse management strategies. The simulation outcomes suggest that, with coordinated interventions (such as habitat corridors, buffer zones, and controlled land-use practices), the frequency of conflicts can be significantly reduced and the connectivity and success of ecosystems and restoration efforts can be enhanced. The findings highlight the application of MAS simulations in the decision-making process of conservation planning, and they help to avoid conflicts and sustainably manage ecosystems. Overall, the article highlights that agent-based modeling can align ecological dynamics with human preferences in problematic socio-ecological settings.

This article presents a comprehensive analysis of the historical evolution of social service centers in Ukraine as key institutions within the national social policy framework. The author traces the transformation of their functions, target groups, organizational forms, and intervention methods from early charitable models to a modern, normatively regulated system of social services. The study identifies five distinct developmental stages: pre-modern (prior to the 20th century), Soviet (1917–1991), post-Soviet (1991–2000), institutionalization (2000–2020), and contemporary (2020–2025). Each stage is examined in terms of its dominant social priorities, beneficiary categories, and methodological approaches. In the pre-modern period, social assistance was primarily religious in nature, grounded in Christian ethics of mercy and care, and implemented through monasteries, parish communities, and charitable brotherhoods. The Soviet era saw the instrumentalization of social work within an ideological framework, emphasizing labor participation and loyalty to the state, with support provided through trade unions, women’s councils, and commissions. The post-Soviet phase marked the emergence of professional social work, notably through the establishment of Youth Social Service Centers and the initial codification of social work as a distinct field. During the institutionalization period, the scope of services expanded, standardized procedures were introduced, and interagency cooperation was strengthened. The contemporary stage is characterized by decentralization, digitalization, and crisis response in the context of war, with social service centers integrated into the National Social Service of Ukraine and focused on supporting internally displaced persons (IDPs), veterans, and trauma-affected populations. The article draws on verified legal acts, strategic documents, academic sources, and historical records to provide a holistic understanding of the institutional development of social service centers as adaptive agents of support for vulnerable populations. The findings are relevant for researchers, practitioners, policymakers, and administrators engaged in the reform and delivery of social services in Ukraine.

In China’s exam-oriented EFL education system, undergraduates face high academic pressure, hindering vocabulary mastery and psychological well-being; this study introduces an adaptive framework to address these challenges. This study presents a groundbreaking approach to improving English as a Foreign Language (EFL) vocabulary mastery and psychological well-being among Chinese undergraduate students by combining an adaptive computer agent-based digital game (ACA-DG) with mind mapping and Runge–Kutta Pairs of Orders 6(5) modeling. Adaptive agents customize interventions to align with students’ learning and emotional requirements, mind mapping structures vocabulary knowledge, and Runge–Kutta methods simulate dynamic learning and psychological processes. A 12-week experimental study at a Chinese university compared an experimental group (ACA-DG with mind mapping, n = 75) to a control group (standard DGBL, n = 75). Vocabulary assessments, motivation, self-efficacy, anxiety, and life satisfaction scales, alongside interviews and observations, revealed a 30.2% vocabulary score improvement, increased motivation, and reduced anxiety in the experimental group. The Runge–Kutta model predicted learning paths with 95% accuracy. This interdisciplinary framework provides innovative tools for EFL educators, merging computational accuracy with emotional support, and advocates for scalable technology-driven learning solutions.

Modern enterprises depend on cloud platforms to support dynamic workloads. Traditional reactive autoscaling mechanisms struggle to maintain performance during sudden spikes or seasonal fluctuations. This research proposes a cloud-native predictive scaling framework that uses machine learning forecasting and reinforcement-based decision policies to allocate resources proactively. The system incorporates multivariate LSTM time-series forecasting, gradient boosting regressors and an adaptive RL-based scaling agent. Testing on a simulated enterprise commerce system demonstrated a 28 percent improvement in response times and a 22 percent reduction in unnecessary resource allocation compared to standard HPA (Horizontal Pod Autoscaler) strategies. The system integrates seamlessly with Kubernetes, enabling container-level scaling with minimal overhead. Stress tests under high-load scenarios confirm the framework’s robustness and cost efficiency. The results indicate that predictive autoscaling is a viable and necessary methodology for next-generation IT infrastructures. Future development will explore multi-cloud orchestration and fault-tolerant scaling.

In recent decades, migration and climate change have emerged as two of the most urgent moral and political challenges confronting the twenty-first century. Predominantly portrayed in mainstream media and political discourse within the Global North as exceptional and extraordinary phenomena, the intersection of migration and climate change is increasingly framed through securitization narratives. This article critically examines the social imaginaries surrounding climate change-induced (im)mobilities, with particular attention to the extent to which these narratives are articulated in terms of justice. Empirically, the study draws on a corpus of 78 fictional stories produced by academics, media professionals, and activists, collected via the “Imaginary wor(l)ds” method during participatory seminars. This creative, practice-led methodology invited participants to collaboratively imagine and narrate future scenarios of climate (im)mobilities, aiming to bridge academic research with public engagement and foster new imaginaries through collective storytelling. The approach is grounded in the recognition that creative writing can disrupt dominant knowledge systems and open discursive space for alternative interpretations. The aim was to reflect on whether an effective collaboration between science, media, and activism might generate counter-narratives for climate (im)mobilities. The stories were analyzed using a typology of frames—victims, security threats, adaptive agents, and political subjects—to identify recurring patterns and the extent to which justice-oriented perspectives emerge. By integrating insights from mobility justice and climate justice frameworks, the study opens a path on how creative, participatory methodologies can reframe public and policy discourses on climate migration, foregrounding justice-oriented approaches over securitized paradigms. The article concludes by underscoring the importance of decolonial and intersectional perspectives in shaping future imaginaries and informing equitable policy responses to global climate (im)mobilities.

Traditional teaching evaluation methods are often retrospective and coarse-grained, lacking the continuous, moment-to-moment feedback required for real-time pedagogical adaptation in language training. Existing intelligent tutoring systems frequently fail to address this, underutilizing multimodal behavioral signals and leaving a critical gap in understanding whether such signals are essential for learning effective policies. This paper introduces a Reinforcement Learning (RL) framework to solve this problem. We propose a hybrid cognitive-linguistic model using a Proximal Policy Optimization (PPO) actor-critic agent, which operates on a novel 516-dimensional state vector that fuses a 512-dimensional semantic embedding from a pre-trained T5 model with a 4-dimensional vector of simulated cognitive-behavioral signals (correctness, response time, attention, hint request). Tested in a simulated learner environment built on the Tatoeba corpus, our agent autonomously discovers a highly effective policy, achieving a mean episodic reward of 6.563, on par with the optimal heuristic baseline. The identified policy, albeit optimal in the simulation, embodies a counterintuitive method that significantly prioritized task repetition by the learner. Our hypothesis is confirmed by a critical ablation experiment: an agent that is deprived of the cognitive-behavioral signals does not learn, and they are as good as a random baseline (mean reward 5.213). The present work gives conclusive evidence that multimodal cognitive-behavioral cues are not only supplementary but are an inevitable part of learning by adaptive pedagogical agents. We mainly provide validation of a hybrid state representation that allows an RL agent to learn effective teaching strategies, making way for more useful and customized educational technologies.

Abstract In the industrial sector, optimizing energy consumption is a critical priority, yet the deployment of effective monitoring systems is often hindered by the need for extensive, last-mile customization. This paper addresses this challenge by presenting a flexible and modular architectural blueprint for an energy submetering system based on IoT devices. The proposed architecture is built on three core principles: the use of non-invasive and plug-and-play sensing technologies, a lightweight and adaptable data collection agent, and a high-performance time-series data store. This modular design allows for incremental deployment, starting small and scaling rapidly without significant operational disruption. We validate the effectiveness of this architecture through a real-world case study in a manufacturing facility, where we monitored five production departments for three months. The analysis of the collected data, visualized as 24-hour heatmaps, immediately revealed key operational patterns, including shift schedules, breaks, and overnight anomalies. Furthermore, we outline several high-impact use cases enabled by this granular data, including energy-aware job scheduling, data-driven sizing of renewable assets, and predictive maintenance. This paper provides a practical guide for Energy Managers, Plant Operators, and industrial decision-makers seeking to implement scalable and cost-effective energy management solutions.

High-risk environments such as emergency response, industrial safety, and critical infrastructure monitoring generate complex, multimodal data streams. Traditional AI systems struggle with the latency demands, incomplete observations, and dynamic conditions present in these domains. This paper proposes an adaptive multimodal agent architecture that integrates sensor fusion, uncertainty modeling, and hierarchical decision layers to support fast, explainable recommendations. The agent combines a transformer-based fusion encoder with a Bayesian uncertainty head and a lightweight policy planner optimized for real-time execution on edge devices. We introduce an adaptive gating mechanism that activates only the relevant modality experts based on data quality and environmental context, significantly reducing computation. Experiments use three public datasets and two simulated emergency scenarios (chemical leak detection and structural collapse risk). Results show that our system improves inference speed by 28–40% while increasing prediction robustness under missing modality conditions. Human evaluators found explanations generated from the uncertainty head clearer and more trustworthy. A final case study demonstrates deployment on a mobile robot in a controlled disaster-response drill. We discuss failure cases and implications for safety certification of AI systems.