Translation from Natural Language to Stock Flow Diagrams

Mental models are bases to recognise phenomena and make plans to improve situations. They can be expressed in model builders' natural language. It is also necessary to examine mental models using a computer simulation. The Computer simulation requires expressions, which can be translated into computer codes. Therefore, model builders need to translate their model from their own natural language to simulation-friendly language, i.e. stock flow diagrams in System Dynamics. It is widely recognised that this translation is sometimes difficult not only for people who are beginners of System Dynamics but also for people who are experienced in the field. This paper discusses a possible translation procedure and shows an application of it. The proposed procedure is designed to use a subset of a natural language as an intermediate language. This idea is applicable regardless of variety of natural language.