Automating Scientific Knowledge Extraction (ASKE)

The goal of this DARPA project is to develop an open-source information extraction pipeline that can be used to ingest epidemiology papers and associated source code repositories, extract entities that represent scientific model concepts (e.g., model parameters, functions, causal relationships, etc.), and use these entities to construct a semantic knowledge graph, that can be traversed to create meta-models (e.g., novel combinations/parameterizations of modeling concepts), and/or probablistically reason about how to go from a set of known input(s) to a set of unknown but abstractly identifiable, unitful output(s).


Scientists construct and analyze computational models to understand the world. That understanding comes from efforts to augment, …