RI: Small: Principles of Normative Multiagent Systems for Decentralized Applications

区块链和物联网等新兴技术为多个自治方（个人或组织）提供了以尊重每个人利益的新方式进行合作的机会。这种考虑也适用于构建涉及多个代理的软件应用程序，但是我们构建这些软件应用程序的方法并没有跟上这些机会。目前大多数方法都是从单个组织的单一角度关注数据和控制。该项目旨在克服多智能体系统设计中的两个概念性挑战：干扰参与者自主性的过度约束交互，以及以对人类没有意义的方式表征交互。这个名为Mia的项目旨在实现创新应用，促进用户的自主性，为用户带来预期的生活质量效益，并为用户和企业带来效率提升。特别是，在不损害严格性的情况下实现权力下放可能会导致实现科学和商业互动的新方法，从而为社会创造价值提供新的途径。该项目旨在通过研究多智能体系统的理论模型和编程技术来支持分散应用程序的自治各方的互操作。Mia框架试图通过指定消息的通信协议对自治各方之间的交互进行建模，以支持完全分散的制定。首先，Mia将研究指定多代理协议的方法，以便协议可以由分散的参与者有效地制定，每个参与者都基于其本地知识和私人决策自主行动，并且一个参与者可以验证另一个参与者对协议的遵从性。其次，Mia将研究自治各方如何通过将规范形式化为定向道义结构来实现其联合活动的一致性。第三，Mia将研究如何引出利益相关者的要求，尊重他们的自主权；从记录的交互中挖掘规范；并系统地从规范模型中归纳出灵活的协议。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Emerging technologies such as blockchain and the Internet of Things offer opportunities for multiple autonomous parties (people or organizations) to collaborate in new ways that respect everyone's interests. This consideration applies as well to building software applications involving multiple agents, however our ways of building these software applications have not kept up with these opportunities. Most current approaches focus on data and control from the unitary perspective of a single organization. This project seeks to overcome two conceptual challenges in multiagent system design: over-constrained interactions that interfere with participants' autonomy, and the characterization of interactions in a manner that is not meaningful to humans. This project, dubbed Mia, aims to enable innovative applications that promote users' autonomy with anticipated benefits in the quality of life for users as well as gains in efficiency for users and enterprises. In particular, enabling decentralization without compromising on rigor could lead to new ways of realizing scientific and business interactions and thereby provide new pathways for value generation in society. This project seeks to support interoperation of autonomous parties for decentralized applications by investigating theoretical models and programming techniques for multiagent systems. The Mia framework seeks to model interactions between autonomous parties via communication protocols that specify messages in support of fully decentralized enactments. First, Mia will investigate ways to specify multiagent protocols such that a protocol can be effectively enacted by decentralized participants, each acting autonomously based on its local knowledge and private decision making, and one participant can verify the compliance of another participant with respect to the protocol. Second, Mia will investigate how autonomous parties may achieve alignment on their joint activities through a formalization of norms modeled as directed deontic constructs. Third, Mia will investigate ways in which to elicit stakeholder requirements that respects their autonomy; mine norms from logged interactions; and systematically induce flexible protocols from a model formulated in terms of norms.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.