Compound Knowledge Graph-Enabled AI Assistant for Accelerated Materials Discovery

Abstract Materials scientists are facing increasingly challenging multi-objective performance requirements to meet the needs of modern systems such as lighter-weight and more fuel-efficient aircraft engines, higher heat oxidation-resistant steam turbines. While so-called second wave statistical machine learning techniques beginning accelerate materials development cycle, most science applications data-deprived when compared vastness complexity search space possible solutions. In line with DARPA’s vision third AI approaches, we believe a combination data-driven domain knowledge will be required achieve true revolution in discovery. To that end, envision have begun reducing practice system fuses three forms knowledge—factual scientific knowledge, physics-based and/or analytical models, expert knowledge—into single ‘Compound Knowledge Graph’ which contextual reasoning adaptation can performed answer complex questions. We this Compound Graph-based nucleus collaborative assistant supports stateful natural language back-and-forth dialogs between discovery new materials. This paper details our vision, summarizes progress date on turbine blade coating use case, outlines thoughts key challenges making reality.