A Quantitative Approach to the Evolution of Domain-Specific Modeling Languages

摘要

Domain-Specific Modeling Languages (DSMLs) are modeling languages that are effective only for particular domains, and DSMLs can make software development much simpler. It is often desirable for DSMLs to be usable repeatedly and for a long time since developing DSMLs is costly and time-consuming. However, this is not easy because it is difficult to evolve existing DSMLs. Conventional evolution of DSMLs is usually conducted based on the information obtained from domain experts (i.e., the interview results of domain experts). However, the DSML problems understood by domain experts can be incomplete, which makes it difficult for DSML developers to judge how to evolve DSMLs and whether an evolution is effective or not. In this paper, we propose an approach to the evolution of DSMLs using quantitative information such as the application data of DSMLs.