A Method of Extracting Synthesis Process from Scientific Literature and Evaluation in the field of All-Solid-State Batteries

Abstract

In the field of inorganic materials, synthesis processes, which are procedures of chemical experiments, are essential for automatic experimental design. However, most material synthesis processes are written in scientific literature as natural language. In this paper, we propose a framework developed by combining a deep learning-based sequence tagger and a simple heuristic rule-based relation extractor. Our experimental results demonstrate that the sequence tagger and rule-based relation extractor can extract flow graphs with high performance on a manually annotated corpus of the scientific literature on all-solid-state batteries.