The VoLL-KI project aims to advance higher education on three levels: macro, meso, and micro. This is achieved by combining data- and knowledge-based artificial intelligence (AI) approaches. Some of the components being developed include intelligent tutoring systems, explainable and interactive machine learning, chatbots, virtual reality, and recommendation systems. Additionally, AI introductory courses for sub-disciplines are being created.

At the macro level, evidence-based advancement of degree programs is being implemented. This is done through context-adaptive, correctable recommendations for individual study planning.

The meso level is focused on learner-specific diagnosis and support in course units. Study progress data is available via an established data warehouse system and is systematically expanded. Data on individual students' existing competences and those that need to be developed are combined with data on specific groups, such as gender and educational biographies, to create customized recommendations for study planning. Students can request explanations for recommendations, explore alternatives and correct premises at any time.

The micro level is expanded through the monitoring of learning and performance trajectories on an individual and group-specific level. This information is integrated into the data warehouse and made available to those responsible for the study programme as a dashboard. The offers developed will be evaluated over the course of the project by means of surveys and logfile analyses in order to optimise them formatively.

**Research focus**

Researchers from the fields of AI, AI-related areas of computer science, computer science didactics and educational research from three neighbouring universities are cooperating on the project. The focus of the project is on the computer science degree programmes at three locations - a large, strongly engineering-oriented computer science, a medium-sized, strongly interdisciplinary computer science and a small, strongly application-oriented computer science. Towards the end of the project and afterwards, the successful components will be extended to other study programmes and the project results will be integrated into the quality management processes of the participating universities.

The aim of the project is to analyse complex holistic business processes with semantic methods of process mining using a hybrid AI approach of machine learning. Explainable artificial intelligence (AI) methods in the form of interactive dashboards are to be used to improve the comprehensibility and transparency of the process models and to realise their refinement and correction through a human-in-the-loop approach. In detail, domain-specific background knowledge from ERP systems and from subject matter experts is to be used and explicitly represented as a knowledge graph. The underlying data model of the event log is to be enriched with semantic meta-information (semantic event log) in order to explicitly represent causal and hierarchical relationships between events and processes and to use them in process reconstruction. Based on machine learning approaches, existing process discovery methods will be adapted and developed as specialised declarative learning methods on the basis of the semantic event log. Finally, the results of the methods will be validated on the basis of business processes of partner companies with real data.