Proposal for an elementary reaction knowledge base for interpreting computational results of chemical reaction processes

Abstract

In recent years, computational chemistry has developed methods to computationally determine detailed processes of chemical reactions. Artificial Force Induced Reaction (AFIR) is one such method that uses quantum computation to explore a wide variety of plausible reaction pathways. However, it is not easy to interpret such complicated reaction paths without additional information. In this study, we propose a system that can interpret reaction processes using simple chemical reactions that form complex chemical reactions, called "elementary reactions", which are used to describe chemical reaction processes in papers and other publications. Ontologies of chemical reactions including elementary reactions, such as RXNO, exist, but their knowledge is mostly descriptive and insufficient for automatic interpretation of reaction processes. Therefore, we propose a new knowledge base framework for elementary reactions that focuses on describing changes in the structural patterns of molecules before and after the reaction. In addition, a method for interpreting reaction processes based on this knowledge base is investigated and validated using actual computational results.