Abstract: Knowledge Query and Manipulation Language (KQML) is a language of typed messages, usually understood as speech-acts, encoded as ASCII strings (in a LISP-like syntax), that are transported over TCP/IP connections, and aimed at knowledge and information exchange between software systems that are viewed as Virtual Knowledge Bases. KQML, which first appeared almost 10 years ago, has come to define the concept of an ACL and in the process the ACL has become the centerpiece of a large category of agent systems. Inevitably an ACL has become a loosely-defined concept that encompasses a variety of issues which may or may not be ACL-relevant depending on one’s point of view. The more “conservative” viewpoint advocates that the semantics of the message types is the one and only real issue. Agent development suggests though that semantics is the least important concern when one actually builds an agent system. The efforts of many researchers to develop multi-agent systems have brought to the foreground issues and considerations that are at least as important as the semantics for interoperable agent systems. After introducing some of the basic concepts relating to Agent Communication Languages, we cover KQML and FIPA ACL, the two existing fully-specified ACLs. We give a brief introduction to their semantics and the issues relating to semantic descriptions of ACLs. We then shift our focus beyond the semantics and point to emerging threads of research in the ACL community. The issues that we deem relevant to the widest possible acceptance of ACLs include alternative syntactic encodings, services and infrastructure, integration with the WWW, and specification of conversation protocols.

Abstract: The Java programming language and technologies have been used extensively in the construction of web-based components and applications specifically in the context of electronic markets (emarkets) development. Java-based components can fulfill a variety of atomic roles in an emarket scenario from graphical user interfaces to complex database functions. However, in distributed settings, a great deal of monolithic code must be wrapped into the core functionality offered by Java components to allow for communication or message passing. Java introduces the Jini specification and associated functions to handle this distributed collaboration. Unfortunately, even Jini adds unnecessary overhead to the core functionality of these atomic components. Consequently, agents can be deployed in conjunction with these Java technologies to assist in communication as well as other nonfunctional concerns. The Knowledge Query and Manipulation Language (KQML) is a protocol that has been used in implementing agent communication. This paper suggests the use of agents to broker communication between atomic components. KOJAC (KQML over Jini for Agent Communication) is an approach to agent-based communication implemented with Jini services, specifically JavaSpace. KOJAC defines methods using Jini services that conform to the KQML protocol and incorporate an object-oriented ontology.

Abstract: In this paper, the basic architecture of an agent is proposed and a federated agent system based on the basic architecture is implemented. The basic architecture consists of a knowledge query and manipulation language (KQML) handler, a knowledge interchange format (KIF) interpreter, and design flow control codes. The KQML handler is a module for managing KQML performatives received from an agent. The KIF interpreter provides the support services of content-based routing, translation using a matching scheme, and simple inference. A Java template package was used for communication between agents. A federated agent system was then applied to a preliminary ship design to show the handling process of the agent system and its applicability in a distributed design environment.

Abstract: The idea of a software entity that performs tasks on behalf of a user across the Internet is now well established. We introduce in this paper a new approach to service discovery and QoS negotiation over the Internet. Our approach presents a framework for service discovery and QoS negotiation at the network level that rely on two concepts: multi-agent systems and agent communication languages (ACL). In this framework, a user and service agents engage in a structured communication through the mediation of a QoS Broker Agent and a Facilitator Agent. Here, the Facilitator Agent acts on behalf of several service agents. It acquires information from these service agents and acts as a single point of contact to supply this information to the User Agent via the QoS Broker Agent. A number of service discovery protocols like the Service Location Protocol (SLP), and Sun Microsystem's Jini has been designed for restricted environments and do not scale to the entire Internet. In order to provide an infrastructure for large scale Internet applications, we designed a prototype multi-agent system that is able to discover resources and negotiate QoS at the network level.

Abstract: The idea of a software entity that performs tasks on behalf of a user across the Internet is now well established. We introduce in this paper a new approach to service discovery and QoS negotiation over the Internet. Our approach presents a framework for service discovery and QoS negotiation at the network level that rely on two concepts: multi-agent systems and agent communication languages (ACL). In this framework, a user and service agents engage in a structured communication through the m ediation of a QoS Broker Agent and a Facilitator Agent. Here, the Facilitator Agent acts on behalf of several service agents. It acquires information from these service agents and acts as a single point of contact to supply this information to the User Agent via the QoS Broker Agent. A number of service discovery protocols like the Service Location Protocol (SLP), and Sun Microsystem's Jini has been designed for restricted environments and do not scale to the entire Internet. In order to pro vide an infrastructure for large scale Internet applications, we designed a prototype multi-agent system that is able to discover resources and negotiate QoS at the network level.

Abstract: The problem addressed by Mediation to Implement Feedback in Training (MIFT) is to customize the feedback from training exercises by exploiting knowledge about the training scenario, training objectives, and specific student/teacher needs. We achieve this by inserting an intelligent mediation layer into the information flow from observations collected during training exercises to the display and user interface. Knowledge about training objectives, scenarios, and tasks is maintained in the mediating layer. A designer constraint is that domain experts must be able to extend mediators by adding domain-specific knowledge that supports additional aggregations, abstractions, and views of the results of training exercises. The MIFT mediation concept is intended to be integrated with existing military training exercise management tools and reduce the cost of developing and maintaining separate feedback and evaluation tools for every training simulator and every set of customer needs. The MIFT Architecture is designed as a set of independently reusable components which interact with each other through standardized formalisms such as the Knowledge Interchange Format (KIF) and Knowledge Query and Manipulation Language (KQML).