The use of machine learning technology is still significantly lower in small and medium sized enterprises than in large enterprises. It seems that there are specific challenges in the implementation of data-driven methods, that hinder SMEs in their adoption. One approach to support the initialization and execution of such methods is the use of boundary objects, e.g., canvases, serving as a visual communication document. As it is not clear which approaches are being pursued in detail and how they are interrelated, in this paper, a systematic literature review is being presented, that identifies and analyzes 18 canvas artifacts. These canvases represent the status quo and they can be grouped into four distinct categories of different foci. The aggregation of the fields and questions provides an essence of canvas contents, to point out gaps and ultimately to expand the canvas approach as well as ML adoption.
(Workshop: Künstliche Intelligenz für kleine und mittlere Unternehmen (KI-KMU 2021)