The Improvement of Sustainable Employability Transfer in Higher Education Institutions Using Large Scale Data Bases and Machine Learning

Abstract

In the previous chapter's analysis, human intelligence had to determine which individual key terms are to be searched for in an existing corpus. This implies that the entire corpus (e.g. the titles of theses) can never be examined with regard to all inherently existing terms at once, but always only for those aspects that are already in the focus of interest. Consequently, this chapter examines software-supported possibilities for automated data processing. A solution for improving the transfer of sustainable employability in higher education institutions based on large-scale databases and machine learning is presented and critically discussed. As an interdisciplinary approach, an AI software developed in the context of research on serious games is adapted and applied to the corpus of theses. This software is based on the Latent Dirichlet Allocation (LDA) topic model, which can be used to classify large amounts of text and derive topics from text corpora. Finally, an indicator is presented that includes aspects of the process of LDA modelling as well as possibilities of additional extensions to decipher time cycles of topics and calculate future trends in order to assess the future importance of a management concept or its disappearance from scientific discourse. From this, the importance and thus the teachability of concepts in the future can be derived, which can be used, for example, for curriculum development or the weighting of the content of subjects at universities.