The CoyPu project resorts to semantic technologies and using AI analysis approaches to develop a data-driven platform to support economic crises. Through the CoyPu framework, macroeconomic, industry-specific, or internal company data can be networked, analyzed and evaluated to enable more efficient crisis management. By providing semantically modeled data in the cloud and flexibly configurable AI analysis tools, the platform will allow for high-quality and up-to-the-minute insights regarding economic facts, trends, impact correlations, and forecasts.
In this subproject of the CoyPu project, components for extraction and data processing and methods for AI-based analyses for trend and risk assessments are developed for the CoyPu platform. The basis for this is a knowledge graph-based semantic data integration method. A contextual knowledge graph will be built and evaluated, and enriched using AI methods based on the vocabularies and ontologies. For this purpose, methods for unstructured data sources as well as semantic integration such as mapping and link generation, link prediction, knowledge graph completion for structured data sources are used on the one hand, and temporal and spatial graph-based neural networks for sequential data on the other hand. These data and knowledge management methods will facilitate data representation at different granularity scales and for the dynamic adaptation of the CoyPu framework to new situations of companies or users.
Additionally, these approaches are supported by automatic machine learning methods, which can automatically adapt to current requirements. Thus, sustainable usability for a wide range of applications and scenarios will be supported. This novel combination of state-of-the-art techniques for knowledge representation using knowledge graphs and automatic machine learning will focus on the explainability of the analysis and prediction results through the integration of symbolic (semantic) and sub-symbolic (statistical) methods.