Towards Distributed Cognitive Expert Systems

The process of Datafication gives rise to ubiquitousness of data. Data-driven approaches may create meaningful insights from the vast volumes of data available to businesses. However, coping with the great volume and variety of data requires improved data analysis methods. Many such methods are dependent on a user's subjective domain knowledge. This dependency leads to a barrier for the use of sophisticated statistical methods, because a user would have to invest a significant amount of labor into the customization of such methods in order to incorporate domain knowledge into them. We argue that machines may efficiently support researchers and analysts even with non-quantitative data once they are equipped with the ability to develop their own subjective domain knowledge in a way that the amount of manual customization is reduced. Our contribution is a design theory - called the Division-of-Labor Framework - for generating and using Experts that can develop domain knowledge.