Big Data meets Big Knowledge: Design Principles for the Combination of Visual Analytics and Knowledge Visualization in Collaborative Business Contexts.

Professionals utilize sophisticated visual analytics tools (such as Tableau, PowerBI, SAS, R, or Python) to collaboratively make sense of data and for evidence-based decision making. For many contexts, however, the quantitative insights made visual in the data need to be combined with qualitative insights based on experience and know-how. Only by taking into account previous experiences, context awareness, and relevant expertise, can the data visualized in business intelligence tools be used adequately. Examples of such data-knowledge combinations include the following constellations: • Risk managers need to add their understanding of market cycles and customer profiles to the risk metrics provided by their risk analysts. • Marketing professionals need to interpret customer data in light of previous experiences and future growth plans. • Strategists need to examine the provided KPIs and dashboards by reflecting on their industry context and the nature of the strategic decisions to be taken. The question thus arises how to best combine big data with big knowledge (in the sense of experience and expertise)? In a three year national science foundation project, we have explored this question, working with real-life organizations (such as a central bank, a travel industry company, or a defense contractor), and we have summarized our findings in a set of design principles. These simple rules instruct programmers, designers, analysts, and managers in how to complement their visual analytics tools with compelling ways of representing relevant expertise graphically. In the paper we describe the scope of our project and the design principles. We provide visual examples that bring the principles to life, and we mention their limitations. We present a human-centered design approach that expands current visual analytics by incorporating knowledge visualization. As a method we have used a design science approach, using the theory of collaborative dimensions by Green et al. and the concept of boundary objects to develop and test prototypes. The results besides the prototypes are design principles that anyone working in the analytics field can use.