Aim/Purpose: This study proposes a hybrid learner profiling framework that combines behavioral data and semantic analysis based on Large Language Models (LLMs) to build interpretable and actionable learner representations. Background: Existing learning analytics approaches rely primarily on quantitative interaction logs or static self-reports, failing to capture the underlying cognitive and emotional reasons behind learner behaviors. Most existing profiling approaches describe what the learner does, but not why they behave in a certain way. Methodology: The framework relies on two complementary pillars. The first analyzes data from virtual learning environments (VLEs). The second uses an LLM as a semantic annotator to extract cognitive, self-regulatory, and emotional dimensions from learners’ textual outputs. The approach was validated using real-world offline datasets and synthetic pedagogical archetypes. Contribution: This work contributes a multidimensional learner representation that bridges the gap between observable interaction logs and unobservable internal states. Findings: The behavioral profiling module achieved a dropout prediction AUC of 0.871. Semantic annotation showed no correlation between cognitive complexity and emotional tone (r = 0.00). In controlled validation, planning cues were detected in 92.3% of valid Strategist messages, compared with 0-16.7% of messages from the other archetypes. Recommendations for Practitioners: Researchers should treat learner profiling as a multidimensional and partially latent construct requiring complementary data sources. Recommendation for Researchers: Researchers should consider learner profiling as a multidimensional and partially latent construct, combining multiple data sources rather than relying solely on easily accessible interaction logs. Impact on Society: This framework helps educational organizations reduce dropout rates and improve the overall quality and inclusivity of digital learning environments. Future Research: Future developments will focus on deploying the framework in real-world settings, such as integrating it into a real Learning Management System (LMS) and evaluating its impact on adaptive learning and personalized support.
Aim/Purpose: This study aims to identify the key factors influencing the adoption of integrated enterprise resource planning and business intelligence systems (ERPBI) in small and medium-sized enterprises (SMEs). Using the technology-organization-environment (TOE) framework, the study develops a conceptual framework explaining how technological, organizational, and environmental determinants shape ERPBI adoption. Background: ERP and BI systems are increasingly interdependent: ERP systems integrate operational data, while BI systems convert that data into decision-support insights. In SMEs, adopting integrated ERPBI systems is especially challenging because of limited resources, technical expertise, and managerial capacity. Since existing studies often examine ERP and BI separately, an integrated, SME-focused framework is needed to better explain the combined adoption of operational and analytical capabilities. Methodology: A two-phase research design was employed. First, a systematic literature review of peer-reviewed studies published between 2017 and 2025 was conducted to extract ERPBI adoption determinants. Second, ten domain specialists were selected through purposive sampling based on their experience in ERP, BI, SME digitalization, or information systems adoption. They evaluated the relevance and importance of the identified determinants, which were then ranked using aggregated mean expert scores. Contribution: This study consolidates fragmented ERP and BI adoption research into an integrated ERPBI perspective and proposes a validated TOE-based conceptual framework specifically tailored to SMEs. Findings: The findings suggest that ERPBI adoption in SMEs is influenced by twelve technological, organizational, and environmental determinants: top management support, training, competitive pressure, communication, firm size, compatibility, relative advantage, perceived usefulness, artificial intelligence tools, big data analytics ability, cloud computing facility, and pandemic recovery planning. Rather than presenting these factors as definitive predictors of adoption success, the study synthesizes prior evidence and expert feedback to propose a structured framework for further empirical investigation. Recommendations for Practitioners: For practitioners, the proposed framework should be viewed as a decision-support guide rather than a prescriptive implementation model. SME managers may use the framework to assess key readiness areas, including technological compatibility, organizational resources, managerial support, employee capability, vendor support, and external pressures. However, the relevance and priority of these factors may vary across SME size, sector, digital maturity, and resource availability. Recommendation for Researchers: Future research should empirically test the proposed framework across different SME sectors and regions and investigate post-adoption outcomes using longitudinal approaches. Impact on Society: Effective ERPBI adoption can strengthen SME resilience, promote sustainable business practices, and contribute to economic stability through enhanced digital transformation. Future Research: Further studies are encouraged to explore emerging factors such as cybersecurity readiness, data governance, and responsible AI practices in ERPBI adoption.
Aim/Purpose: The primary objective is to address the structural operational cost crisis in modern SOCs by designing and formally specifying a three-layer agentic AI framework that integrates semantic alert triage, adaptive reinforcement-learning response, and episodic knowledge synthesis into a unified architecture. Background: Modern SOCs are experiencing an acute operational crisis. Exponential growth in alert volume, high false-positive rates (>40%), and chronic analyst attrition have created a perfect storm. Existing rule-based SIEM and single-agent SOAR approaches only achieve 20–55% alert automation and fail to address the full operational lifecycle. Methodology: The framework evaluation is based on a structured comparison with seven benchmark systems across four dimensions (threat coverage breadth, integration completeness, cost quantification, and adversarial safeguards), each rated on a five-level ordinal scale using replicable criteria. Contribution: The paper makes four contributions: (1) a formal algorithmic specification of a three-layer agentic AI architecture including three pseudocode procedures and a PPO state-action-reward formalism; (2) a quantitative operational cost projection framework explicitly distinguishing designed targets from measured performance, with a maximum designed workload reduction of 86%; (3) a systematic four-dimensional comparative analysis against seven benchmark frameworks demonstrating that CyberAgent is the only framework achieving full architectural completeness (integrating all three of semantic triage, adaptive RL response, and episodic knowledge synthesis simultaneously), an architectural claim requiring empirical confirmation; and (4) dual adversarial safeguards (prompt injection mitigation and reasoning consistency verification) absent from all seven benchmark frameworks. Findings: CyberAgent is a theoretical design-science artefact that has neither been implemented nor empirically evaluated. All quantitative projections are designed to achieve targets grounded in prior work, not verified outcomes: an alert automation rate of 85–90%, an analyst workload reduction of 56–86% (design target 86% under the product-rule independence assumption), an MTTR reduction of 65–75%, and a false-positive reduction of 60–70%. These projections require empirical validation using CybORG and the DARPA OpTC dataset, with this as the primary future work priority. Recommendations for Practitioners: The PTL’s Chain-of-Thought (CoT) reasoning traces provide human-readable decision narratives that enhance transparency and may support auditability workflows relevant to GDPR Article 33, HIPAA, PCI-DSS, and SOX. However, CoT traces do not automatically satisfy regulatory auditability or compliance requirements; they are one architectural input to a broader compliance process. Formal legal and compliance assessment by qualified professionals is required before deployment in regulated environments. Practitioners should treat CoT output as decision-support documentation, not as regulatory certification. Recommendation for Researchers: Future work should also develop federated DRL training protocols to ensure policy convergence under data-scarce conditions and rigorously test the adversarial robustness of the PTL’s consistency-checking mechanism against novel prompt-injection strategies. Impact on Society: If empirically validated, CyberAgent could make enterprise-grade cyber defence more accessible to mid-market organisations that cannot afford 24/7 SOC analyst staffing, by substantially reducing the alert-triage workload. All such impact claims are conditional on validation results. Future Research: Priority directions include: (1) empirical implementation and red-team validation across diverse enterprise environments; (2) federated DRL training to address the data sharing constraints that limit policy learning in regulated sectors; (3) extension of the KSL to support cross-organizational threat intelligence sharing; and (4) longitudinal studies measuring analyst skill development and human-AI trust calibration under progressively increasing levels of CyberAgent autonomy.
Aim/Purpose: Individual differences in signers’ speed and amplitude cause variations in keypoint sequences, leading to redundant features that degrade the classification performance of sign language recognition models. Background: To reduce the impact of individual differences among signers, this paper proposes a Dynamic Adaptive Representation Calibration (DARC) module that helps the model focus on the most informative features. Methodology: The study utilizes component-based modeling of high-dimensional semantic features and introduces sample-adaptive learnable weights. The method was validated using large-scale sign language datasets, specifically WLASL2000 and MSASL1000. Contribution: The proposed DARC module dynamically adjusts the importance of different features for each input, allowing the model to focus more on informative patterns and reducing the influence of individual differences among signers. Findings: The proposed method achieved a Top-1 accuracy of 56.86% on the WLASL2000 dataset (a 3.2% relative improvement over the state-of-the-art DSTA-SLR) and 65.88% on the MSASL1000 dataset. Visualizations confirm that the proposed module significantly improves feature discriminability and clustering clarity. Additionally, multi-stream training validations demonstrate consistent performance improvements across spatial and temporal dynamic streams, effectively raising the model’s overall representation ceiling. Recommendations for Practitioners: Developers of practical sign language recognition systems should consider integrating dynamic adaptive calibration to better handle natural variations among different signers in real-world environments. Recommendation for Researchers: Researchers should further explore dynamic feature recalibration techniques to improve model robustness against individual variations in sequence-based recognition tasks. Impact on Society: By improving the accuracy of keypoint-based computer vision models, this research helps develop more accessible and scalable tools to bridge the communication gap between the deaf/hard-of-hearing community and hearing individuals. Future Research: Future studies should explore the application of the DARC module to other video-based action recognition tasks or investigate its performance in real-time continuous sign language translation.
Aim/Purpose: The use of machine learning models in graduate employability prediction is constrained by limited transparency. This limitation weakens their value for organizational decision-making in higher education. Background: An Explainable Artificial Intelligence (XAI) framework is proposed to fill this void. Using this framework, the predictive output generated will be translated into insights that are interpretable and useful for sensemaking and evidence-based management. Methodology: The experiments were conducted on a database containing 225,384 graduate records. The best accuracy results were obtained using the CatBoost algorithm, and global and local explanations were generated using SHAP and ICE. Contribution: Predictive accuracy is integrated with interpretability in this study. A framework is produced as a result, designed to support accountable, data-driven decision-making processes within higher education institutions. Findings: Education level and language proficiency are identified as dominant predictors of employability outcomes. Substantial influence is also exerted by GPA, university ranking, and internship experience, while demographic attributes such as gender make a negligible contribution. Both aggregate feature importance and heterogeneous individual effects are exposed through the explainability analysis. Recommendations for Practitioners: Institutions should prioritize language development, targeted academic support, and expanded internship pathways. Interpretable analytics should also be embedded in curriculum planning and policymaking. Recommendation for Researchers: Explainable AI should be examined within broader educational landscapes in future research. Additional variables should be incorporated, with particular attention directed toward fairness, bias, and model robustness. Impact on Society: The results indicate potential for more equitable and effective policy making. They also align with broader objectives related to workforce readiness and Sustainable Development Goals on education and employment. Future Research: Future work should rely on longitudinal and multi-institutional data. The practical impact of explainable AI in real decision-support environments also requires systematic evaluation.
