Sixty Years - and more - of Data Modelling

Data (conceptual, data, information, knowledge) modelling is still the work of an artisan, i.e. an art in the best case, made by humans, because of the need for human intelligence. Data modelling is an essential part of Information System (IS) design, specifying how data is implemented as part of an IS. The principles of data modelling follow the evolution of IS development paradigms, and these in turn follow the progress of technological changes in computing. Although technology has changed a lot during the decades of commercial use of computers – since the early 1950s to now, close to 70 years – data modelling is still based on the same basic principles as decades ago. Or is it really so? Finding the answer to this question was the main motivation to start writing this paper. Since the future is more interesting than the past, we set our research problem to be “What are the challenges for data modelling in the future?”. The reason for this is that we see some significant changes in the future in the data modelling sector which we wanted to examine. However, the future is a continuum of the past. The future cannot be fully understood without understanding the past. Humans also tend to forget the details of the past. Even the most remarkable innovations from the past have become part of the new normal. Consequently, at the beginning of our paper we look shortly at the progress of data modelling during the era of commercial computing. Our focus is on the recent past and we look at the technological changes that have been of key importance in data modelling in the role of triggers and enablers. To find the answer to our research question, we retrieved some recent studies handling the future of data modelling and analyse the challenges found in these sources. The paper is concluded by some future paradigms. In general, the big changes seem to be the growing importance of Artificial Intelligence (AI) and machine learning (ML) as its fuel. AI not only conducts algorithmic rule-based routines, it has learning capability, which makes it more intelligent and adaptable, and able to compete with human intelligence, even in data management tasks.