News of November 25, 2022

Success for ICCL and TU Dresden: CPEC Funded for Four More Years

TU Dresden has achieved an outstanding success. Together with partners from Saarland University, the Max Planck Institute for Informatics, and the Max Planck Institute for Software Systems we may continue doing research in the Transregional Collaborative Research Centre 248 Foundations of Perspicuous Computing (CPEC) on comprehending the cyber-physical world respecting human interaction. The German Research Foundation (DFG) provides funds for 14 projects inside CPEC for four more years, starting in January 2023.

CPEC is fundamentally about building systems that can be understood by humans. A first step towards that is generating explications for results obtained from a software system. For a system based on logical inferences, proof trees provide a valuable source of explications. But truly understandable (i.e., “perspicuous”) systems need to go even further and have proper explanations that are designed to be consumed and understood by the human user. For instance, typical proof trees can be so large that the human user can no longer recognise meaningful patterns. In phase 1 of CPEC, ICCL members Franz Baader and Stefan Borgwardt have jointly explored feasible explications for description logics (DL) in project A3. Furthermore, Markus Krötzsch has investigated fundamenental aspects regarding expressivity of rule languages and also provides usable APIs as well as front-ends for rule reasoners within the first phase of CPEC.

For the second phase of CPEC, Markus Krötzsch joins the team of principle investigators in A3. New is the focus on DLs with arithmetic. "We are also trying to adapt some of the techniques developed for explications of DLs in phase 1 for dealing with rules" says Markus. The main challenges are that rules may have higher arities than 2, making proof trees wider. Furthermore, the recursive nature of rule languages also tends to make proof trees deeper. Additionally, rule languages are investigated for providing insights (via explications) in developing software (focus of project E3). The targeted consumer of such explanations is the software developer.