High-Utility Itemset Mining (HUIM) incorporates both the profit and profitability factors, a critical component of data mining. Nevertheless, the majority of HUIM algorithms are developed to operate on a single machine, which proves inefficient for big data due to limited memory and processing resources. This paper proposes a novel Coverage, Unit profit utility-based High Utility patterns from Certain data (CUHUC) framework. The proposed framework follows two stages: High Utility Mining and Pruning, with a novel thresholding strategy. In high utility mining, the uncertain dataset is converted into a certain dataset. Certain factors like Unit profit, Purchase quantity, and Coverage-based High Utility Itemset (UPC-HUI) are generated from the dataset. Further, the generated list is then pruned to eliminate unpromising candidates. The remaining promising items in a fixed order are arrangeds. An innovative Improved Pruning with Novel Thresholding Strategy (IPNTS) is proposed to remove the unpromising candidates. This can set a minimum utility value to filter the significant pattern and propose the Enhanced Secretary Bird Optimisation (ESBO) algorithm to obtain the optimal minimum utility. The ESBO scheme attained a runtime of 42.412 s, memory usage of 196 MB, and cost rate of 0.49500, signifying improved performance compared to conventional methods.
Artificial Intelligence (AI) has emerged as one of the most sought-after academic disciplines worldwide, primarily due to its strong career prospects and expanding industrial applications. This study provides data-driven career guidance for prospective AI graduates by analysing recent labour market trends and examining the influence of geographic location on AI career opportunities. A structured dataset was analysed using a Random Forest regression model to forecast market trends and employment trajectories. Projections from the World Economic Forum's 2025 report on emerging jobs towards 2030 were incorporated to provide a forward-looking perspective on AI employment dynamics. The findings indicate substantial growth in AI-related occupations, with net job growth for AI and Machine Learning roles projected to increase by approximately 82% by 2030. Among the most promising career paths are Data Scientists (+113% projected growth), Machine Learning Specialists (+82%), Software Engineers (+57%), Cybersecurity Specialists (+53%) and Robotics Engineers (+40%). These results offer evidence-based insights to support informed academic and professional decision-making for students pursuing AI careers. These trends provide empirical support for strategic academic and workforce planning in AI-related fields. Students are encouraged to align their research and skill development with these domains, as AI is poised to become a central component of the future workforce.
Digital transformation has been brought out as one of the most critical strategic initiatives by organisations that are keen on making innovation and remaining competitive in technology-driven environments. In the banking industry, the gradual adoption of digital technologies has completely transformed the service provision processes, operational procedures, and the decision-making processes of managers. This paper examines how digital transformation affects innovation in Jordanian commercial banks, focusing specifically on the three key aspects, namely, technological infrastructure, legislative and policy frameworks, and digital skill sets. A descriptive-analytical research design was adopted and the data were collected using a structured questionnaire that was offered to both the senior and middle managers of the commercial banks in Jordan. The number of valid responses was 124 and was analysed statistically, including a reliability test, Descriptive statistics, a normality test and regression analysis. The current analysis indicates that the digital transformation has a statistically significant positive impact on innovation in the environment of commercial banking in Jordan. Among the dimensions that were examined, legislative and policy frameworks were the most powerful drivers, followed by technological infrastructure and digital competencies. The data also depict average levels of adoption of digital transformation interventions and practices that accompany them, hence pointing to significant improvement opportunities. The given empirical study adds to the body of existing literature on the topic of digital transformation and innovation by providing primary data on the banking industry of a developing economy. The derived insights have practical implications for banking executives and policy-makers in the society because of the importance of enabling regulatory environments, investment in digital infrastructures, and the long-term development of the digital capabilities of employees to foster innovation.
Short-term electricity demand forecasting is essential for operational planning in electricity systems, particularly in cooling-dominated regions. While prior studies have largely focussed on improving predictive accuracy, less attention has been given to how forecasting models can be structured and evaluated for operational decision support. This study proposes a structured framework that connects high-frequency data inputs, interpretable feature engineering and machine learning-based short-term forecasting with downstream decision and governance considerations. The forecasting component of the proposed framework is empirically evaluated using high-resolution electricity load and weather data from the Tetouan (Morocco) dataset, incorporating temporal, lagged and climate-derived features. The proposed framework, instantiated through boosting-based and other regression methods in its forecasting layer, demonstrates reliable short-term predictive performance and provides a foundation for operational planning and governance in smart energy systems. The framework is intended to be transferable to smart-meter-based electricity systems in hot and arid regions.
Digital technology has become integral to language learning, offering interactive tools and resources. Online platforms and language learning apps provide accessibility and convenience for learners. Integration of digital tools enhances engagement and facilitates personalised language acquisition. By employing a Technology Acceptance Model (TAM) framework, this research investigates high school students in the Chennai district’s user acceptance of integrating digital technology in Foreign Language Learning (FLL). The research employs a random sampling method to gather data from a diverse group of high school students. The independent variables considered in this investigation include Technology Self-Efficacy (TSE), Learners’ Motivation (LM) and Learners’ Behaviour (LB). Perceived Usefulness (PU) and Perceived Ease of Use (PE) are conceptualised as mediating variables, which subsequently influence learners’ behavioural intention to use digital technology in FLL. The TAM serves as the theoretical framework to analyse and understand the relation between these variables. The findings indicate that Technology Self-Efficacy (TSE), Learners’ Motivation (LM) and Learners’ Behaviour (LB) have a noteworthy influence on PU and PE. Mediation results show that PU and PE significantly mediate the effects of TSE, LM and LB on students’ intention to use digital technology, with PU having a stronger role than PE. By delving into the perspectives and experiences of high school students, this study seeks to uncover insights that can inform educators, policymakers and technology developers on enhancing the effectiveness of integrating digital technology into FLL environments. The research findings seek to enhance the overall conversation on the acceptance and utilisation of digital tools in educational environments.
Seven large language models were compared against binary vulnerability classification using the DiverseVul benchmark: StarCoder2-7B, Phi-3.5-Mini-Instruct, DeepSeek-Coder-6.7B-Instruct, Llama3-8B-Instruct, Gemma-7B-IT, Qwen2.5-7B-Instruct and GPT-4o-mini. Each model was evaluated using a unified zero-shot prompting protocol with generation settings chosen to minimise stochasticity, and API-accessible models run with temperature = 0 or the lowest value allowed by the platform, while locally executed models used greedy decoding. On a balanced test set of 3,000 samples, accuracy and precision, recall, F1-score and Cohen Kappa were calculated. Accuracy ranged from 39.93% to 56.23%. The highest accuracy (56.23) and the highest Kappa (0.1246) were obtained with GPT-4o-mini. Using standard interpretation guidelines, a Kappa of 0.1246 indicates only slight agreement, suggesting that the apparent performance reflected by accuracy is weaker than it initially seems. The code-specialised models were more prone to over-flag vulnerabilities and accumulate false positives, whereas instructions-tuned general-purpose models were less prone to do so. Confusion-matrix profiles and inference-time measures are also provided. There is no introduction of a new method of detection. The paper therefore controlled zero-shot evaluation of these models without fine-tuning, or in a task-specific manner.
This study aimed to analyse how the Academic and Support Faculties perceive the performance of Management School graduate students. A qualitative study design was embraced to examine the perceptions of Academic and Support Faculties on various aspects of teaching and learning. Results revealed variation with regard to teaching-learning processes held by faculty members. Other members of the faculties appeared content with status quo andragogy methods, yet some urged the inclusion of more interactive and participatory teaching methods. Additionally, respondents emphasised the necessity to focus on student performance and offer essential support to enable them to achieve academic objectives. Based on these results, it has been proposed that the institution (AC) should incorporate greater elements of active teaching in its graduate programs, including problem-based learning, case studies, and simulations. This integration is expected to create parity between student expectations and needs, which subsequently facilitates overall learning among graduate students.
Security Operations Centres (SOC) are increasingly encountering more alerts, overstretched analysts, and delays in responding due to the growing complexity of the infrastructure as well as the generation of cyber threats. Even though automation and SOAR platforms have brought greater levels of stability in operations, they are more of a rule-based system and rely on human judgment. New advances in artificial intelligence (AI) allow now transitioning to agentic SOC architecture, where intelligent systems identify, emphasise, and take response actions independently with minimal human oversight. In this paper, it is hypothesised that a multidimensional assessment framework with an Efficiency, Effectiveness, and Experience (3E) model can be created to consider how agentic AI has been used operationally and organisationally in SOC worlds. On the basis of imitative yet analytically valid security telemetry data obtained within a period of 90 days of simulated observation, we perform a comparative evaluation of the traditional SOC workflows, which were concerned with analysts, and the workflows based on AI that was improved. A qualitative analysis of analyst workload distribution and decision control is a complement to quantitative indicators like MTTD, MTTR, false warning frequency, and alert escalation frequency. They are found to demonstrate that agentic AI will greatly decrease the time of detection and response, stabilise operational performance with a high load of alerts, and convert the roles of the analyst to controlling management. Besides enhancing efficiency, the findings indicate a structural change in the functioning of SOCs, and autonomy is one of the aspects in the future responsibility of SOCs. The research furnishes an operational guide to assessment and furnishes strategic direction to those organisations that are advancing to intelligent security operations.
The rapid integration of Generative AI (GenAI) in higher education has triggered a paradigm shift from initial disruption to widespread normalisation, yet the specific nature of student engagement with these tools remains underexplored. Based on the adapted Extended Technology Acceptance Model (TAM3), this study aims to explore and investigate whether university students utilise GenAI primarily as “Task Implementors” to automate academic workload or as “Learning Catalysts” to deepen cognitive understanding. A quantitative survey was given to 433 undergraduate and graduate students in three leading universities one in Canada and two in Egypt. Cluster analysis identified a dominant group of “balanced integrators” who pragmatically leverage AI automation to complement and perfect their work, rather than replacing it. The results further demonstrate that GenAI adoption is mainly driven by perceived suitability (functional utility) rather than desirability (hedonic enjoyment), confirming that students view these tools as essential professional instruments rather than entertainment. These findings suggest that traditional restrictive policies are rendered obsolete by the practicality of AI technology, necessitating a shift toward process-oriented assessment strategies that acknowledge AI’s permanent role in the academic ecosystem. The study findings have implications for the future of curriculum design, pedagogical strategy, developing assessment and institutional policymaking in higher education.
Employee attrition significantly threatens organisational knowledge retention and competitive performance. This study proposes a machine learning framework for predicting employee attrition, grounded within knowledge management and decision support theory. The framework integrates intelligent data preprocessing, feature engineering, SMOTETomek class balancing, ensemble learning and stratified five-fold cross-validation to ensure rigorous, leakage-free performance estimation. All proposed models significantly outperformed the logistic regression baseline across all metrics (paired t-test, p < 0.01). The Deep Learning model achieved the highest accuracy (0.898) and recall (0.548), while the Voting Ensemble achieved the highest AUC-ROC (0.882), representing an 84% improvement in recall over baseline. Feature importance analysis identified income, age, tenure and manager relationships as chief attrition predictors. The proposed framework equips HR practitioners with actionable decision support tools to proactively manage talent and preserve organisational knowledge.
The need for Arabic Dialect Identification (ADI) is increasing because of the large amount of spoken language in Arabic-speaking countries. While most studies have focused on written dialects using Arabic Script, very little has been done to identify the dialect in Romanised Arabic (Arabizi), which contains many irregular spellings, varied vocabulary, and mixed language (code-switching). In this study, we attempt to address this problem through developing a hybrid approach using both character-level Convolutional Neural Networks (CNNs) and embeddings based on transformers that were fine-tuned on Arabizi data. Additionally, we developed a normalisation process to handle differences in orthography. Our experiments with several social media datasets showed our proposed system had an F1-score of 0.87 compared with other approaches at 6% higher. We provide additional resources, including annotations for these datasets, and offer some methodological guidelines for working with Romanised dialectal Arabic. Finally, we describe specific applications of this work, such as sentiment analysis and machine translation for content created by younger speakers and migrant communities.
Software ecosystems generate complex data traces which can be represented through graph models and stored in graph databases. While large language models (LLMs) offer promising no-code interfaces for querying such databases by translating natural language to database queries, their effectiveness heavily depends on how database schema information is provided. This paper presents a comprehensive evaluation of different schema injection techniques for LLM-driven graph database query generation. We systematically evaluate four key dimensions: schema source, prompt placement, query categories, and results relevancy. Our automated evaluation pipeline records execution outcomes across 50 test queries spanning five query categories. In addition, a human evaluation is conducted for the returned data from each query execution. Results show that LLM-generation of Cypher queries is sensitive to schema source and injection point. In addition, results show that providing the dynamically generated schema through the APOC.meta.graph method yields more relevant and executable queries on average compared to static schema summaries, while injecting schema information in the system prompt outperforms injection in the user prompt. The evaluation provides actionable insights for optimising LLM-driven graph database interfaces and highlights the need for advanced context and software engineering to enable reliable LLM-based solutions.
This paper analyses recent scientific output on resilience in the fields of management and business. Based on a bibliometric analysis of publications indexed in the Scopus database for the period 2014-2023, and using VOSviewer software, the study provides a map of the research landscape on resilience. The analysis covers 1,924 documents published in 160 journals by 5,335 authors affiliated with institutions in 148 countries. The results highlight the structure of the literature and the main research trends, offering useful guidance for future work. This study contributes to the literature by providing one of the few systematic reviews of resilience based on bibliometric analysis and scientific visualisation.
Background: Social media usage (SMU) has grown in popularity among young people, especially in Generation Z (Gen Z). Communicating with peers, gaining new skills and information fast, and maintaining relationships with adult mentors like parents, relatives and instructors are some of its advantages. Objective: This study aimed to investigate how social comparison and cyberbullying interact to influence emotional well-being, providing new insights into the psychological risks associated with social media usage among Gen Z. Methods: An online survey of 531 Indian social media users was carried out between January and March 2024 among 531 Indian social media users (adolescents and young adults aged 12 to 27) enlisted via convenience sampling. Structural equation modelling (SEM) using SPSS and AMOS was employed to evaluate the proposed hypotheses. Findings: The study confirmed a significant direct positive effect of social media use on emotional well-being (b [Formula: see text] 0.184, [Formula: see text]) and a significant indirect effect mediated by social comparison ([Formula: see text]). Cyberbullying emerges as a critical moderator variable, dampening the positive links between social media use (SU) and social comparison (SC) ([Formula: see text]) and between social media use and emotional well-being (EW) ([Formula: see text]). These quantitative results elucidate the nuanced influence of social dynamics on the emotional well-being of Indian Gen Z social media users.
Autonomous artificial intelligence (AI) agents are no longer advisory. They execute transactions, orchestrate multi-agent pipelines and modify enterprise data with minimal human intervention. Yet the governance frameworks organisations rely on were built for a different artefact: one that recommends rather than acts. This paper argues that delegation, not architecture, is the primary variable that governance frameworks for agentic AI must address and that existing frameworks, including NIST AI RMF and the EU AI Act, do not adequately operationalise governance at the delegation level. A multi-corpus bibliometric analysis of 795 peer-reviewed publications (2020-2026) provides evidence of three structurally isolated scholarly communities (AI ethics governance, MLOps operationalisation and agentic enterprise integration) developing in parallel without convergence, leaving the high-autonomy, high-accountability quadrant theoretically underserved. Grounded in three complementary theoretical pillars (IT Governance theory, Socio-Technical Systems theory and IS Artefact Delegation theory), we derive the Delegated Agentic Governance Model (DAGM): a conditional governance matrix that assigns governance requirements to each level of autonomy delegated to AI agents across three tiers (Advisory, Operational, Autonomous). We introduce Generative AI governance debt as a prerequisite construct, articulate seven design principles and derive three falsifiable propositions linking delegation-governance alignment to enterprise failure rates. The DAGM provides AI managers with an immediately actionable governance readiness instrument and establishes the theoretical foundation for a research agenda on delegation-calibrated AI governance across finance, healthcare and manufacturing.
Osteoporosis is a rare bone disease leading to loss of bone tissue and mass. Osteoporosis disease makes the bones weak by reducing the bone strength and leads to fractures. Osteoporosis commonly affects middle-aged women. Osteoporosis disease is diagnosed using Dual Energy X-ray Absorptiometry (DEXA) as it helps in defining the bone mineral density; however, the process is expensive. DEXA has set its margin as the golden standard in osteoporosis disease diagnosis. The latest research performed in osteoporosis disease to verify the effectualness and analyse different diagnosis factors through clinical validation is limited. Therefore, this work provides a survey on the diagnosis of osteoporosis risk using deep learning techniques and also focuses on exploring the different applications-oriented technologies defined in the medical field for eliminating osteoporosis risk. This survey investigates how deep learning methods enhance the early detection and prediction of osteoporosis, a condition characterised by low bone density and increased fracture risk. Deep learning techniques have shown promising results in various medical applications, including image analysis and disease diagnosis. By leveraging these techniques, the survey helps in developing a model that can analyse bone density scans and other relevant data in real time to assess the risk of osteoporosis accurately. The survey delves into the details of deep learner training on large datasets of bone scans, patient information and other medical data to enable the automated detection of osteoporosis risk factors. Furthermore, it explores the challenges and opportunities associated with implementing deep learning techniques in osteoporosis risk assessment. In addition, it assesses the potential benefits of incorporating additional data modalities, such as genetic information or lifestyle factors, into the deep learning model to enhance the overall predictive power for osteoporosis risk assessment. By exploring the fusion of diverse data sources through deep learning frameworks, the survey aims to uncover synergies that lead to more comprehensive and personalised risk evaluations. Finally, this research endeavours to pave the way for more effective and personalised healthcare interventions in the field of bone health. The results show that the CNN achieved an accuracy of 99.23% and the MCNN attained 99% accuracy leading to faster and more accurate diagnosis and personalised treatment plans in osteoporosis diagnosis, which permits for quicker mediation when required and reduces the possible difficulties.
Background: Text summarisation is a procedure that shortens the original source material and separates out important details. Moreover, multi-document text summarisation through integrating the abstraction and extraction approaches is incomplete and remains a major difficult research issue. However, effectively learning the exact meaning of the summarised text is still hard. Therefore, it is significant to address the complications of the standard multi-document text summarisation models. Aim: A novel framework is introduced by employing deep learning approaches. In the beginning, the required text data employed for the validation are collected from the internet resources. Methodology: Initially, the pre-processed text is offered to the text summarisation stage. In this phase, the developed framework utilised Syntax-aware Region Attention-based Bidirectional Encoder Representations from Transformers Integrated with Adaptive Dilated Residual Long Short-Term Memory (SR-B-ADRLNet). Moreover, the parameter of SRA-BERT-I-ADRLSTM is tuned using an Enhanced Uniform Random Variable-based Waterwheel Plant Algorithm (EURV-WPA) for attaining the multi-document text summarisation outcomes. Results: From the validation results, the cosine similarity rate of the developed multi-summarisation model is 0.6205. Conclusion: Thus, the developed multi-document summarisation model using a deep learning model is an efficient tool that produces an informative and succinct summary from a source material.
Unsupervised clustering has become an attractive solution to personalise the learning with AI-assistance in English without the necessity of labelled results. Standard similarity metrics tend to assess all features of learners in the same way, restricting their capacity to reflect pedagogically significant variations. This work is a continuation that builds on a baseline cosine-similarity clustering model, adding a weighted cosine similarity structure-based clustering framework that enhances the discovery of the persona of learners in learning the English language with the help of artificial intelligence. With the relevant large-scale dataset of 15,000 learner interactions with tools like ChatGPT, Grammarly, Duolingo and ELSA Speak, we experimentally consider various schemes of weighting, where the primary concern is that of learning gain, task engagement and error reduction. The feature weighting that is proposed combines with the common standardised preprocessing and dimensionality reduction feature and allows more expressive similarity modelling with no complexity increase to the model. Numerous clustering methods are compared with the use of weighted and unweighted cosine similarity on the Silhouette score, Davies-Bouldin index and Calinski-Harabasz index, finding significant and consistent improvements on both Euclidean and unweighted cosine baselines, with the ability to produce Silhouette scores up to 0.77 and well-separate and pedagogically interpretable clusters of learners. The refined clusters have a more distinct differentiation in the learning efficiency and task-engagement profiles, which makes them even more appropriate to be used in downstream personalisation and recommendation tasks. Placed in a non-native learning of English in the GCC context, the work contributes to unsupervised learner modelling and gives a solid methodological basis for future adaptive tutoring and chatbot-assisted learning systems.
In the modern artificial intelligence-driven banking regime, the world is witnessing exponential growth of transactional and behavioural data, and therefore, for establishing and sustaining financial institutions, it is imperative that higher-order customer risk management and creditworthiness evaluation are done. This requires not only developing knowledge architecture but also establishing decision-aware threshold optimisation. This study proposes a unified model that combines customer segmentation with creditworthiness profiling based on multi-view clustering applied together with predictive risk scores. The proposed framework uses the gradient boosting decision tree (LightGBM) classifier to identify the level of customer credit risk using a combination of consolidated demographic, financial and behavioural variables. This algorithm is commonly used in financial risk analytics because it has demonstrated good performance in dealing with nonlinear relationship modelling and in dealing with disproportionate structured banking data. The empirical analysis is based on a large real-world dataset. It comprises 1,013,193 transactions, 2,000 customers and 6,146 card payments. It also captures detailed transactional, behavioural, card-level and demographic characteristics. The representations of the customers are built on three complementary views, namely (i) demographic and financial health (income, debt, age, credit score); (ii) card and credit utilisation; and (iii) transactional behaviour and expenditure dynamics. The multi-view embedding strategy and unsupervised clustering resulted in two separate groups of customers bordering the distinction between a high-value cohort and a higher-risk cohort. The accuracy of the baseline model is 86.5% and its ROC-AUC is 0.935, which highlights its discriminatory effectiveness. Further optimisation through decision-threshold optimisation, improves operational performance and eventually yields the highest F1-score of 0.843, balanced precision of 0.838 and recall of 0.848. This analysis also indicates that the model can be implemented on various risk considerations; it is scalable and precisely guides risk-conscious financial decision-making, reconciling more advanced machine learning technologies with viable banking and credit-management needs.
Job satisfaction plays a significant role in employees’ performance and commitment. In the service sectors, such as airports, employees interact with customers on a daily basis. Thus, their satisfaction impacts the organisation’s performance. This study examines the impact of hygiene (work environment, job benefits, and interpersonal relationships) and motivators (professional development and recognition) on the job satisfaction of customer service employees at Auckland Airport. The study respondents included both male and female participants of different ages and socio-economic groups. The target sample population was selected through purposive sampling, and data from about 126 respondents were collected through a questionnaire. Descriptive analysis was used to examine the impact of hygiene factors and motivators on the job satisfaction of customer service employees at Auckland Airport. The study found that recognition in the work environment had a significant influence on job satisfaction; job benefits and interpersonal relationships had a significantly lower influence on job satisfaction, and professional development also presented a limited influence on job satisfaction.
