The ITAVER5O project develops sophisticated computational architectures within a metaverse environment, integrating advanced machine learning (ML) models, synthetic data, generative AI, edge computing, and a patented live-action interactivity feature. The core objective is to support high school students in their educational orientation and career planning, offering an immersive, controlled audiovisual setting based on predictive ML-based computational psychology algorithms. Through this gamified experience, Italian students will engage in various activities that will help profile their curricular and behavioral traits, providing valuable insights for personalized educational guidance. Following an initial literature review of the decision-making models identified by contemporary cognitive science, the project will implement a collection of decision-making data from two distinct groups: high school students (final two years) and university students (first two years). The comprehensive data collection effort aims to capture a wide array of student behaviors, preferences, and decision-making processes, essential for training robust ML models. The high school data, being unlabelled, will be used to model general behavioral patterns without predefined categories, whilst university student data will serve as a benchmark to evaluate the predictive model's accuracy, offering insights into the effectiveness of their educational choices and experiences. After collecting and comparing data between these two identified samples, the project will proceed with the generation of synthetic data. To enhance and enrich the dataset, ITAVER5O will employ advanced generative techniques, including oversampling and deep-learning models such as Variational Autoencoders (VAEs) and Generative Adversarial Networks (GANs). These techniques generate synthetic data that mirrors the training set, addressing data imbalances and ensuring a comprehensive dataset for robust ML model training. The project will then design the predictive model: utilizing exploratory data analysis, the project identifies relevant features and applies supervised ML techniques to design an accurate model aimed at forecasting students' educational trajectories and outcomes, leveraging the enriched dataset for higher precision and reliability. The subsequent design of the ITAVER5O gaming platform will feature three meta-planets, representing the three knowledge domains according to the ERC taxonomy. This controlled environment will facilitate the profiling of students' curricular and behavioral patterns as they navigate and interact within the metaverse. The platform's design incorporates live-action interactivity, enhancing engagement and providing a dynamic game-based learning experience. The ITAVER5O platform will be tested and deployed using advanced 5G architectures, ensuring low-latency and high-quality connectivity to deliver seamless, real-time interactions while maintaining the immersive nature of the metaverse environment. Following deployment, the platform will be released and promoted by a renowned influencer professor with 4 million social followers, ensuring the refinement of the predictive model with vast amounts of data from the student population.

Keywords: Computational psychology, decision-making models, generative AI, synthetic content, machine learning, live-action interactivity, metaverse, 5G architectures.