
The group Software and Scientific Computing is working on the development of algorithms and software tools for the fast retrieval and exploration of knowledge in structured and unstructured freely available sources. When browsing scientific literature, searching databases or browsing online media, one often asks oneself "Can this be true?" or "What is actually the current state of knowledge?". If you use portals to search the web, you have to search through long lists of results. We are doing research on distributed information systems that are supposed to answer such questions ad hoc. This goes far beyond keyword-based searches. Our data center integrates both structured databases (for example, on proteins, chemicals, active ingredients, clinical studies) and huge unstructured document collections (for example, research articles, patents, package inserts). The goal is to link different sources via the recognition and normalization of concepts and their relationships to highly complex knowledge graphs. To this end, we use modern methods of information extraction to automatically find designations of concepts (including synonyms and abbreviations) using terminologies and ontologies and to put them into relation (relation mining). The knowledge collected in this way is stored in comprehensive graph databases and can thus be used for searches by experts from various biomedical fields (for example, biomedicine, pharmacy, chemistry, biotechnology). We take advantage of modern, big data architectures, open source software solutions (such as Kubernetes, Apache Spark, Apache UIMA, Apache Spring, REACT) and interfaces (such as OpenAPI, OAuth), which we then adapt and extend to suit our needs.