Artificial society is the mapping and abstraction of large-scale complex systems (especially social systems), which has numerous complex characteristics such as diversity and dynamic evolution. It is also the basic part of ACP approach (Artificial societies for modelling, Computations experiments for analysis, Parallel execution for control, which is a methodology in social computing). Thus, artificial society modelling is an important and meaningful work. However, there are still no standardized and accepted approaches to modelling artificial society. With the analysis of artificial society modelling and based-on social organization metaphors, this paper presented Artificial Society Modelling Language-ASML to simplify and standardize processes of artificial societies modelling for ACP approach, including meta-model (including entities and relationships), graphical notations, models (including role model, interaction model, etc.) and so forth. This modelling language could support massive objects modelling, multi-level modelling, social aspects modelling, dynamic evolution modelling, and it has characteristics of high abstraction, multi-viewpoints, easy-to-use and platform-independent. Then, with a case study, the validity of this language has been verified.