The behavior of e-commerce agents can be defined at different levels of abstraction. A formalism allowing to define them in terms of their economic activities, Utility State Machines, has been proposed. Due to its high level of abstraction, this formalism focuses on describing the economic goals rather on how they are achieved. Though this approach is suitable to specify the objectives of e-commerce agents, as well as to construct formal analysis methodologies, this framework is not suitable to define the strategic behavior of agents. In this paper we develop a new formalism to explicitly define the strategic behavior of agents in a modular way. In particular, we reinterpret the role of utility functions, already used in USMs in a more restrictive manner, so that they define strategic preferences and activities of agents. We apply the formalism to define the agents in a benchmark e-commerce agent environment, the Supply Chain Management Game. Since the strategic behavior of agents is located in a specific part of the formalism, different strategies can be easily considered, which enhances the reusability of the proposed specification.
展开▼
机译：可以在不同的抽象级别定义电子商务代理的行为。已经提出了一种形式主义，可以根据它们的经济活动对其进行定义，即Utility State Machines。由于其高度抽象，这种形式主义着重于描述经济目标，而不是实现这些目标的方式。尽管此方法适合于指定电子商务代理商的目标以及构建形式化分析方法，但此框架不适合定义代理商的战略行为。在本文中，我们开发了一种新的形式主义，以模块化方式明确定义代理商的战略行为。特别是，我们以更加严格的方式重新解释了USM中已经使用的效用功能的角色，以便它们定义了代理的战略偏好和活动。我们采用形式主义在基准电子商务代理商环境（供应链管理游戏）中定义代理商。由于代理的策略行为位于形式主义的特定部分，因此可以轻松考虑不同的策略，从而增强了所提出规范的可重用性。
展开▼
