Hypermedia APIs enable the design of reusable hypermedia clients that discover and exploit affordances on the Web. However, the reusability of such clients remains limited since they cannot plan and reason about interaction. This paper provides a conceptual bridge between hypermedia-driven affordance exploitation on the Web and methods for representing and reasoning about actions that have been extensively explored for Multi-Agent Systems (MAS) and, more broadly, Artificial Intelligence. We build on concepts and methods from Affordance Theory and Human-Computer Interaction that support interaction efficiency in open and evolvable environments to introduce signifiers as a first-class abstraction in Web-based MAS: Signifiers are designed with respect to the agent-environment context of their usage and enable agents with heterogeneous abilities to act and to reason about action. We define a formal model for the contextual exposure of signifiers in hypermedia environments that aims to drive affordance exploitation. We demonstrate our approach with a prototypical Web-based MAS where two agents with different reasoning abilities proactively discover how to interact with their environment by perceiving only the signifiers that fit their abilities. We show that signifier exposure can be inherently managed based on the dynamic agent-environment context towards facilitating effective and efficient interactions on the Web.

翻译：超媒体API能够设计可复用的超媒体客户端，这些客户端能够发现并利用万维网上的可供性。然而，此类客户端的可复用性仍然有限，因为它们无法对交互进行规划与推理。本文在超媒体驱动的万维网可供性利用与多智能体系统（MAS）及更广泛的人工智能领域中已深入探索的动作表示与推理方法之间建立概念桥梁。我们借鉴可供性理论与人类-计算机交互中支持开放可演化环境下交互效率的概念与方法，在基于万维网的MAS中引入符示符作为一级抽象：符示符基于其使用的智能体-环境上下文进行设计，使具有异质能力的智能体能够执行动作并推理动作。我们定义了一个形式化模型，用于在超媒体环境中上下文性地暴露符示符，旨在驱动可供性利用。我们通过一个原型化的基于万维网的MAS演示了该方法，其中两个具有不同推理能力的智能体通过仅感知符合其能力的符示符，主动发现如何与其环境交互。我们证明，符示符的暴露可以基于动态的智能体-环境上下文内在地进行管理，从而促进万维网上高效且有效的交互。