Abstract:

The technological advancements of the past two decades have had the potential to enhance educational offerings, such as university lectures. However, many universities still rely exclusively on PDF documents when developing and disseminating teaching materials. While easily accessible, these documents limit interactive and interconnected learning opportunities that digital platforms could offer. Instructors often struggle with effectively digitizing content and enhancing it pedagogically. Educators and content designers frequently lack the expertise in the subject needed to apply their concepts, particularly in challenging fields such as advanced mathematics. Interdisciplinary trained experts are rare.

This contribution introduces a pragmatic model aimed at transferring teaching content into structured knowledge databases, a model of the learning material that categorizes and prioritizes knowledge while enabling individual (and individually chosen) learning paths through intelligent connections and recommendations. This approach is particularly beneficial for large courses, distance learning, or self-learning students. It facilitates innovative teaching concepts such as the flipped classroom and enhances maintainability. And it effectively exploits the potential of digital content in a learning platform or course management system without demanding unrealistic resources.

Motivation:
In the early days of the internet, interconnected knowledge through "hyperlinking" was revolutionary. Today, we are exploring a new way of connecting content: neural networks create intelligent, individualized learning paths. They analyze individual learning styles, track learning performance, and recommend suitable exercises. However, complex technology, lack of explainability, university policies, budget and time constraints, and privacy concerns make such systems challenging and sometimes undesirable in academic teaching. Instead, we revisit the idea of hyperlinking, enrich it with neural network concepts, and map it onto the content structure, not onto the tutor. This way we mimic the brain’s natural learning process: through the repeated occurrence of facts and other learning items, relevant information is marked, and neural connections are strengthened.

Approach:
In our model, the sources of a learning environment are the lecture text, the learning objectives, and ideally, the teaching expert with in-depth subject knowledge. The first step is to analyze the learning content with respect to the learning objectives. Important aspects will be highlighted, presented more frequently, made more accessible, and linked with other knowledge data more often. In a second step, we incorporate elements of the 4C/ID model, such as just-in-time information in tests and scaffolding in holistic task settings, for non-linear competency-oriented structuring of content. The knowledge database is pedagogically supported by creating tasks adapted to the learning taxonomies to ensure that the learning objectives are fully covered.

Based on the learning objectives and an initial script we thus develop a framework that structures content and takes – during the learning process - cross-connections as well as required competencies into account.

Keywords:

Knowledge Databases, content digitalization, individual learning paths.