Veranstaltungsprogramm

<p>The basic concept of personalized learning has become a primary focus within the realm of intelligent and smart education systems as artificial intelligence continues to evolve (Almasri et al., 2020). This review analyzes the current scenario of personalized learning using cognitive diagnosis, learning analytics, and sentiment analysis using the MOOC and Twitter dataset to underline its prime importance in present-day educational frameworks (Khosravi et al., 2020). Personalized learning is advocated for its capacity to calibrate educational experiences to individualized skills and learning progressions. This is facilitated through the utilization of cutting-edge technology and comprehensive insights into educational mechanisms, thereby fostering adaptability and operational efficiency across diverse learning environments (Radovanović et al., 2023).</p>

<p>In this contribution, a comprehensive review of personalized recommendation algorithms is conducted, with a focus on platforms such as MOOCCourse, MOOCCube, EdX, and Coursera. The objective is to explore how these algorithms not only cater to individual learner preferences but also dynamically adapt to their evolving educational needs. The study is centered around multifaceted research on personalized learning, aiming to synthesize various perspectives to provide clear definitions, insights into precise objectives, and relevant educational theories. Additionally, it critically evaluates the influence of technical frameworks on personalized learning outcomes, highlighting their effectiveness in enhancing individual capabilities and addressing specific educational requirements (Shi et al., 2022). Moreover, a detailed survey of the data applications and assessment metrics used within personalized learning is outlined, which constructs an AI framework for data analysis and outcome evaluation (Lu et al., 2021). This framework is deemed essential for guiding future research in the field.</p>

<p>The analysis demonstrates that advanced technologies like Neural Collaborative Filtering (NCF), GPT, and hybrid Transformer models significantly enhance learning systems' adaptability and efficiency (Zhang et al., 2023). These technologies refine student models and recommendation systems tailored to individual learning needs. The study discusses integrating data application strategies and outcome evaluation metrics essential for advancing personalized learning research. NCF's cognitive analysis uses metrics like precision, recall, and AUC-ROC, while GPT models use BLEU and ROUGE scores for natural language processing performance. Personalized learning recommendations are evaluated with ranking metrics such as mean reciprocal rank (MRR) and mean average precision (MAP). The review identifies challenges and future directions for personalized learning, proposing solutions to advance the field (Radovanović et al., 2023).</p>

As educational institutions strive to benefit from the rapid developments in artificial intelligence applications in their teaching and learning practices, they still need systematic guidance in effective utilization of these technologies for educational purposes while carefully addressing users’ and organizations’ needs and adequately overcoming possible challenges and limitations. This paper proposes a comprehensive framework for integrating generative artificial intelligence into digital learning and online education. The Framework for Integrating Generative Artificial Intelligence in Digital Learning (GAIDL) consists of six key components that cover various aspects of AI governance and policy, curriculum, teaching, learning, assessment, ethics, technology, student support, professional development, and evaluation. Each component has specific areas of focus and objectives. The framework aims to help educational institutions leverage the potential of AI to enhance learning outcomes while ensuring ethical and responsible use of AI.

<p>In contemporary educational paradigms, the integration of Large Language Models (LLMs) and advanced Natural Language Processing (NLP) techniques offers a transformative opportunity to redefine personalized learning and mentorship. This contribution presents a framework for the development of Trustworthy AI-Mentors, leveraging LLMs and core NLP methodologies to deliver more secure and personalized learning assistance. The central research question guiding this study is: How can AI-driven mentorship systems effectively support personalized educational experiences while ensuring stringent data privacy?</p>

<p>Traditional educational mentorship models face challenges in addressing diverse learner needs while maintaining data privacy (Jiang et al., 2022; Li &amp; Wang, 2021). The Trustworthy AI-Mentors framework integrates differential privacy, deep learning, and LLMs to create a secure, scalable learning environment. It combines differential privacy techniques with LLMs for large-scale model training while protecting data confidentiality (Tramèr et al., 2020; Papernot et al., 2018). Techniques such as noise addition, data obfuscation, and secure multi-party computation ensure privacy during model parameter or statistic sharing (Wei et al., 2020). Transformer-based LLMs enhance natural language understanding, enabling AI-mentors to provide context-aware, personalized learning through advanced NLP techniques like named entity recognition, sentiment analysis, and dependency parsing (Vaswani et al., 2017; Brown et al., 2020; Radford et al., 2019; Liu et al., 2019). </p>

<p>To evaluate the effectiveness, efficiency, and security of Trustworthy AI-Mentors, a mixed-methods research methodology is employed. This includes quantitative analysis using traditional machine learning metrics such as accuracy, error rates, F1 score, precision, and recall, as well as language-specific metrics like perplexity and BLEU score for assessing LLMs (Kumar et al., 2021). Privacy-preserving metrics, including the privacy loss bound, epsilon, and data utility, are used to ensure data privacy (Jayaraman &amp; Evans, 2019). Additionally, computational efficiency metrics such as training time and inference latency are considered. Qualitative analysis involves user feedback and case studies to assess the user experience and real-world applicability of the AI-mentors.</p>

<p>The adoption of Trustworthy AI-Mentors has significant implications for the future of education, democratizing access to high-quality mentorship and learning resources (Zawacki-Richter et al., 2019). By offering secure, personalized learning experiences, this framework empowers learners while protecting their sensitive information. Additionally, the scalable nature of differential privacy and LLMs ensures the deployment of Trustworthy AI-Mentors across diverse educational contexts, transcending geographical boundaries and socio-economic barriers (Whittlestone et al., 2019; Morley et al., 2021).</p>