Introducing Knowledge Graphs for Hierarchical Time Series Reconciliation

AIKG-SD 2025 Summer School co-located with the NFDI4DS Conference 2025 November 25-26, 2025, Berlin, Germany Introducing Knowledge Graphs for Hierarchical Time Series Reconciliation Dirk Beinert,1,2 Timm Bönke,1 Christian Menden,3 and Michael Martin2 1DATEV eG, 2TU Chemnitz, 3TH Würzburg-Schweinfurt Abstract A recently published real-time system of business indicators based on German SME company data offers new possibilities to nowand forecast the economy in detail and different levels of aggregation (sectors, regions, company size). The data contains taxes, wages and business evaluation data processed and filed monthly by tax consultants displaying a major sample size within Germany (Beinert et al 2025). Recent findings on the optimization of forecasting methods, especially when based on hierarchical data developed more insights and accuracy through the different levels of time series aggregations (Athanasopoulos et al, 2024). Knowledge Graphs (KGs) are crucial using artificial intelligence and especially questionanswering (Yucheng et al, 2020). However, their value within other disciplines than computer science e.g. economics and forecasting only slowly emerges (Tilly and Livan, 2021). This presentation tries to combine all three aspects into a common cognitive map for the purposes of high frequency economic data, its advanced forecasting possibilities and a deeper understanding of how detailed large scale business data helps to analyze the health status of companies and the relevant business sectors. It may open the door for simulations and modelling which might be superior to the well-known and recognized survey-based business climate estimations. The poster will focus on KG construction (Mo et al, 2025) and VAR model training and presents first results. References Beinert, D., Bönke, T., and Menden, C., (2025), Tracking the Economic Performance of Germany‘s SMEs in Real-Time. WIBF 2025. Stock, J.H., and Watson, M.W., (2010), Dynamic Factor Models. Oxford Handbook of Economic Forecasting. Kolassa, Rostami-Tabar and Siemsen (2024), Demand Forecasting for Executives and Professionals. CRC Press. Tilly, S., and Livan, G., (2021), Macroeconomic Forecasting with statistically validated Knowledge Graphs. Expert Systems with Applications. Yucheng. Y, Yue, P, Guanhua, H., and Weinan, E, (2020), The Knowledge Graph for Macroeconomic analysis with Alternative Big data. arXiv.2010.05172. Athanasopoulos, Hyndman, Kourentzes and Panagiotelis (2024), Forecast Reconciliation: A Review. International Journal of Forecasting 2025. Mo, B., Yu, K., Kazdan, J., Mpala, P., Yu, L., Kanatsoulis, C. I., and Koyejo, S., (2025), Extracting Knowledge Graphs from Plain Text with Language Models. NeuroIPS Conference 2025.