Ontologies for fault-tolerant multi-agent systems

本课题研究的重点是多智能体系统的互操作性。多智能体系统的互操作性意味着智能体之间应该能够相互理解。每个代理都有一种精神状态，在这种状态下它存储了正常运行所需的信息。因此，代理必须能够在他们的精神状态之间交换信息。然而，在每个代理中，信息不能以相同的方式描述。因此，工程师应该参考相同的词汇表，以便将收到的信息与他们自己对该信息的描述进行映射。这个词汇表称为本体论。本体是用于引用在特定领域中操纵的概念的词汇表。此外，可能存在几个多代理系统，其中存在必须相互交互的代理。如果为每个多智能体系统定义了至少一个本体，那么我们就面临着集成与多个多智能体系统相关的多个本体的问题。因此，在多智能体系统需要互操作性的背景下，本研究计划提出：1)基于智能体所处的上下文、智能体必须响应的事件以及智能体必须执行的动作，定义一种方法来为多智能体系统开发本体，2)通过识别将被映射的概念以及它们将如何被映射来开发新的本体映射技术，3)开发将容易识别本体的公共和不同概念的方法，4)提出新的算法来合并本体并在合并本体时检测不一致。

This research program focuses on multi-agent systems interoperability. Multi-agent systems interoperability means that agents should be able to understand each other. Each agent has a mental state in which it stores the information required for its well functioning. Hence, agents must be able to exchange information between their mental states. However, information could not be described by the same way in each agent. Consequently, agents should refer to a same vocabulary in order to map received information with their own description of that information. This vocabulary is called an ontology. An ontology is a vocabulary used to refer to concepts manipulated in a particular domain. Furthermore, there could be several multi-agent systems in which there are agents which have to interact with each other. If there is at least one ontology defined for each multi-agent system, then we face the problem of integrating several ontologies related to several multi-agent systems. Consequently, in the context where multi-agent systems require interoperability, this research program proposes: 1)to define a methodology to develop ontologies for multi-agent systems based on the context in which agents are, the events to which the agents must respond, and the actions the agents has to perform, 2) to develop new techniques for ontologies mapping by identifying the concepts that will be mapped, and how they will be mapped, 3)to develop approaches which will ease the identification of common and different concepts of ontologies, 4)to propose new algorithms for merging ontologies and detecting inconsistencies when merging ontologies.