Knowledge graph support for descriptive business analytics

Abstract Business analytics provides decision-makers with the fundamental for an informed, fact-driven choice of the best course of action for an organization. Analysis results, however, are often not self-explanatory, nor is the best course of action following the analysis results always obvious. In order to interpret analysis results correctly, decision-makers require a deeper understanding—knowledge—of the development and analysis process as well as the employed data. When the required knowledge is not properly documented or only possessed by certain individuals, obtaining such tacit knowledge retrospectively can be challenging and costly for an organization. In the worst case, obtaining tacit knowledge may even have become impossible if, for example, the employee with the required knowledge has already left the organization. Even if tacit knowledge about the analysis is indeed documented, another challenge is to retrieve the required knowledge to argue a decision without having to search through large amounts of documented knowledge manually. Building on years of practical experience of an industry specialist in business intelligence and analytics, we propose a method for employing a knowledge graph to capture tacitly available knowledge that is generated during the execution of the business analytics process. The resulting knowledge graph can be queried to provide information about provenance, preprocessing steps, and other characteristics of the analyzed data to support decision-makers with the best possible foundation to correctly interpret analysis results.