The derivation process in rewriting systems, especially in Chomskyan grammars, can be seen as consisting of four stages: (1) selection of the production to apply from a supply of productions available at that step, (2) check of applicability of the production, possibly on different matches, (3) selection of the match at which to apply the production, (4) actual application. In this paper, we explore the different ways in which models of regulated and distributed rewriting have extended these four basic mechanisms and we propose a unifying perspective on them in terms of interaction between two agents, expressed through a particular form of conditional rewriting systems. This introduces a compositional view of derivation processes, where different components, playing specific roles and exhibiting specific behaviours, can be made to communicate and combined into different models of derivation. Existing models of regulated and distributed rewriting can thus be reformulated in this new framework, and new ones can be obtained.

重写系统中的推导过程，尤其是乔姆斯基语法中的推导过程，可以看作由四个阶段组成：（1）从该步骤可用的产品供应中选择要应用的产品，（2）检查产品的适用性， （3）选择要应用产品的匹配项，（4）实际应用。在本文中，我们探讨了规范和分布式重写模型扩展了这四个基本机制的不同方式，并就通过两种形式的条件重写系统表示的两个代理之间的相互作用提出了一种统一的观点。这引入了派生过程的组合视图，其中不同的组件扮演特定的角色并表现出特定的行为，可以进行沟通，并组合成不同的推导模型。因此，可以在此新框架中重新构造现有的规范和分布式重写模型，并可以获得新模型。