語意網路是目前熱門的研究領域。它以現有的網路為基礎，為其內容增添語意，組成網路知識本體。因此，網路上的應用程式可以藉由知識本體達到跨平台運算的目的。另一方面，代理人是一門提昇軟體抽象化的理論。它使軟體由物件典範移轉到智慧運算層次，將軟體模塑成一個具有心智的個體。我們的研究便是將代理人技術應用在語意網路上，藉由代理人讓語意網路應用程式具有智慧化的運算。另一方面，代理人也可以藉由語意網路上的知識本體，實現代理人平台之間的資訊整合。 我們提出代理人在語意網路上的溝通架構，代理人藉由知識本體來運算，因此，我們為溝通架構定義了一套代理人溝通知識本體，藉由這一套知識本體我們可以將代理人的溝通移植至語意網路。我們定義了相對應於每一個代理人溝通階層架構的知識本體以描述階層內的子元件以及階層之間的元件。不同的是，我們也提出一個新的代理人溝通階層，我們在訊息傳遞層與內容語言層之間插入一個代理人心智層，(1) 這樣可以讓訊息傳遞元件可以傳送代理人的心智以朝向到代理人平台之間開放的心智傳遞前進。(2)將心智表示法與內容語言表示法分開，讓心智態度可以不被某些語言所限制，以利分享。 我們的架構嘗試著與語意網路上的程式互動，如：語意網路服務與語意網路應用程式。這些過程與成果都在本文中所描述。 The Semantics Web is one of the popular research areas these days. It is based on the current Web, and adds more semantics to it, on purpose of building the Ontology of Web content. So, application program on Web can make the purpose of cross-platform calculation come true by taking advantage of Ontology. On the other side of coin, agent is a theory able to enhance abstraction of software itself. This theory makes software shift from object paradigm to intelligence agent paradigm, and thus constructs the software into an object with mental model. The purpose of our study is to apply agent technique to Semantics Web, and our method is to make application program of Semantic Web be able to make intelligent computing by utilizing agents. Moreover, agents can realize the information integration among platforms of agents by making capital of Ontology on Semantics Web. This study brings up the communication architecture on the Semantic Web; precisely speaking, agents make computing with Ontology, and we define an agent’s communication ontology for this communication framework. We can transplant agent’s communication to Semantics Web by using this Ontology. Our researches define corresponding ontologies for all agent communication layers for the sake of describing ways of construction of subcomponents inside layers and of components among layers. On the contrary, we want to propose a new agent mental layer. We insert it between the message transport layer and the content language layer in the cause of: (1) Making any message transport component be capable of transmitting mental attitudes of agents, which leads to the openness of mental attitude conveyance amid platforms, (2) Separating the content representation from the mental representation, which can cause mental attitudes not to be limited by certain languages and thus facilitate sharing. Our architecture attempts to interact with a popular the application of Semantic Web such as Semantic Web Service and Semantic Web Application. The experience and effort are also described in this paper.
