Automatisierte Modellierung von akademischen Wissensdomänen als Methode zum innovativen Wissensmanagement

Abstract

The knowledge content of academic knowledge domains is subject to constant change. Modelling these rapidly evolving domains of knowledge is a challenge due to the complexity, constant adaptations, and large amounts of data. This article presents a new epistemological approach to domain modelling based on the theory of human mental modelling. It allows to map these dynamic domains of knowledge. This approach extends the existing technological spectrum of knowledge engineering by a new class of methods. It closes the conceptual gap in the transfer between machine knowledge processing and human use. Human knowledge is characterized by a contextdependent networking of information. These networks are associative in nature. According to this basic functionality of human knowledge, the validated text analysis program T-MITOCAR (Text Model Inspection Trace of Concepts and Relations) analyses the semantic knowledge structures contained in text in a computer-linguistic way and maps them as association networks (maps of knowledge). The resulting knowledge maps resemble a mind map in their appearance. The aim is to provide an adequate re-representation of human knowledge on the basis of text. In a second step, these methods can also aggregate entire text corpora into a map of the knowledge domain. Using these technologies, a complete academic knowledge domain can be modeled automatically as a knowledge map. Such knowledge domain models can be used as instruments of innovative knowledge and information management by institutions with fast-moving knowledge domains, such as universities and specialized information services. The methods enable libraries, for example, to automatically generate a knowledge map from different text corpora, to compare them and to identify overlapping contents of the underlying knowledge domains. Changes in the knowledge domain can be analyzed in an overview-like manner and literature recommendations can be issued based on what has previously been written.