Abstract:

In a world in which information is collected increasingly quickly and comprehensively, it is important to produce high quality didactic material of new topics for lecturing and teaching. Nevertheless, it is virtually impossible for scientists to write textbooks, that meet the expectations of students and the requirements of lecturers, in a timely manner. The learning process itself requires in general a stepwise, hermeneutic approach: Understanding of complex concepts requires the knowledge of the respective individual components. Once the original concept is fully understood, it can be considered as a single building block for even more complex contexts. This process goes on and on and is valid for almost all disciplines. Textbooks must take this into account.

In the area of Information Technology (IT) this approach is called modularity. A module is made up of other modules and is used to create ever larger modules. Abstraction layers serve to keep the ever-increasing complexity bearable for humans. Subject of knowledge changes with each step of abstraction.

The question is how methods of Machine Supported Teaching (MST) can be used to build up abstraction layers automatically. In the long run, abstraction layers could be used not only to support lecturers and teachers in their work, but also to automatically generate textbooks for extravagant or particular difficult areas of knowledge.

In this paper, we focus on generating concrete abstraction layers exemplarily in the field of elementary chess endgames. Even simple questions raise a lot of technical difficulties to overcome. Automatic classification of difficulty levels is decisive. Starting with elementary positions, the level of difficulty is gradually increased by introducing abstraction layers. For humans it is necessary to give the abstraction layers a name so that the associated concept can be memorised more easily. The machines are not able to produce appropriate terms by now. Therefore, also future approaches to MST will require human intervention. Nevertheless, computers can provide valuable services to human teachers.

After discussing numerous general examples, special cases that entail a significant increase in complexity are also dealt with. This is followed by a comprehensive evaluation of the abstraction layer approach.

In the outlook, it is shown that there is still a long way to go before textbooks can be produced completely automatically. Excellent textbooks can be found in many subject areas today. However, they often lag behind new findings for quite some time. Moreover, they do not cover all areas of human interest.

Especially in these areas, automated MST algorithms can make a valuable contribution to learners and teachers.

Keywords:

Abstraction Layers, Machine Supported Teaching, Machine Learning.