Yannis Labrou

Agent communication languages: the current landscape

Despite the substantial number of multiagent systems that use an agent communication language, the dust has not yet settled over the ACL landscape. Although semantic specification issues have monopolized the debate, other important pragmatic issues must be resolved quickly if ACLs are to support the development of robust agent systems. We introduce some concepts useful in discussing agent communication languages and then compare and evaluate the two major ACLs.

A declarative approach to business rules in contracts: courteous logic programs in XML

We address why, and especially how, to represent business rules in e-commerce contracts. By contracts, we mean descriptions of goods and services offered or sought, including ancillary agreements detailing terms of a deal. We observe that rules are useful in contracts to represent conditional relationships, e.g., in terms& conditions, service provisions, and surrounding business processes, and we illustrate this point with several examples. We analyze requirements (desiderata) for representing such rules in contracts. The requirements include: declarative semantics so as to enable shared understanding and interoperability; prioritized conflict handling so as to enable modular updating/revision; ease of parsing; integration into WWW-world software engineering; direct executability; and computational tractability. We give a representational approach that consists of two novel aspects. First, we give a new fundamental knowledge representation formalism: a generalized version of Courteous Logic Programs (CLP), which expressively extends declarative ordinary logic programs (OLP) to include prioritized conflict handling, thus enabling modularity in specifying and revising rule-sets. Our approach to implementing CLP is a courteous compiler that transforms any CLP into a semantically equivalent OLP with moderate, tractable computational overhead. Second, we give a new XML encoding of CLP, called Business Rules Markup Language (BRML), suitable for interchange between heterogeneous commercial rule languages. BRML can also express a broad subset of ANSI-draft Knowledge Interchange Format (KIF) which overlaps with CLP. Our new approach, unlike previous approaches, provides not only declarative semantics but also prioritized conflict handling, ease of parsing, and integration into WWW-world software engineering. We argue that this new approach meets the overall requirements to a greater extent than any of the previous approaches, including than KIF, the leading previous declarative approach. We have implemented both aspects of our approach; a free alpha prototype called Common-Rules was released on the Web in July of 1999, at http://alphaworks.ibm.com.

Evaluation of KQML as an agent communication language

This chapter discusses the desirable features of languages and protocols for communication among intelligent information agents. These desiderata are divided into seven categories: form, content, semantics, implementation, networking, environment, and reliability. The Knowledge Query and Manipulation Language (KQML) is a new language and protocol for exchanging information and knowledge. This work is part of a larger effort, the ARPA Knowledge Sharing Effort, which is aimed at developing techniques and methodologies for building large-scale knowledge bases that are sharable and reusable. KQML is both a message format and a message-handling protocol to support run-time knowledge sharing among agents. KQML is described and evaluated as an agent communication language relative to the desiderata.

An Approach to Using XML and a Rule-Based Content Language with an Agent Communication Language

We argue for an XML encoding of FIPA Agent Communication Language (ACL), and give an alpha version of it, called Agent Communication Markup Language (ACML), which we have implemented. The XML approach facilitates: (a) developing/maintaining parsers, integrating with WWW-world software engineering, and (b) the enriching capability to (hyper-)link to ontologies and other extra information. The XML approach applies similarly to KQML as well. Motivated by the importance of the content language aspect of agent communication, we focus in particular on business rules as a form of content that is important in e-commerce applications such as bidding negotiations. A leading candidate content language for business rules is Knowledge Interchange Format (KIF), which is currently in the ANSI standards committee process. We observe several major practical shortcomings of KIF as a content language for business rules in e-commerce. We argue instead for a knowledge representation (KR) approach based on Courteous Logic Programs (CLP) that overcomes several of KIF’s representational limitations, and argue for this CLP approach, e.g., for its logical non-monotonicity and its computational practicality. CLP is a previous KR that expressively extends declarative ordinary logic programs cf. Prolog; it includes negation-as-failure plus prioritized conflict handling. We argue for an XML encoding of business rules content, and give an alpha version of it, called Business Rules Markup Language (BRML), which we have implemented. BRML can express both CLP and a subset of KIF (i.e., of first-order logic) that overlaps with CLP. BRML expressively both extends and complements KIF. The overall advantages of an XML approach to content language are similar to those for the XML approach to ACL, and indeed complements the latter since content is carried within ACL messages. We have implemented both ACML and BRML/CLP; a free alpha prototype of BRML/CLP, called IBM CommonRules, was released on the Web in July of 1999.

Using Colored Petri Nets for Conversation Modeling

Conversations are a useful means of structuring communicative interactions among agents. The value of a conversation-based approach is largely determined by the conversational model it uses. Finite State Machines, used heavily to date for this purpose, are not sufficient for complex agent interactions requiring a notion of concurrency. We propose the use of Colored Petri Nets as a model underlying a language for conversation specification. This carries the relative simplicity and graphical representation of the former approach, along with greater expressive power and support for concurrency. The construction of such a language, Protolingua, is currently being investigated within the framework of the Jackal agent development environment. In this paper, we explore the use of Colored Petri Nets in modeling agent communicative interaction.

Standardizing Agent Communication

An Agent Communication Language (ACL) is a collection of speech-act-like message types, with agreed-upon semantics, which facilitate the knowledge and information exchange between software agents. From Knowledge Query and Manipulation Language (KQML) to FIPA ACL, ACLs have been a cornerstone for the development of systems of communicating agents, and simultaneously they have been the subject of intensive standardization efforts.Standardizations goal is usability. As a result, although the initial focus on ACLs revolved around establishing the semantics of ACLs, a variety of usability-related questions have entered the picture of standardizing communication among agents. In this article, we present these questions and the work that addresses them, alongside the historical evolution of ACLs, their semantics and the results of their standardization.

Agent-based approach for manufacturing integration: The ciimplex experience

The production management system used by most manufacturers today consists of disconnected planning and execution processes and lacks the support for interoperability and collaboration needed for enterprise-wide integration. This situation often prevents the manufacturer from fully exploring market opportunities in a timely fashion. To address this problem, we are exploring an agent-based approach to intelligent enterprise integration. In this approach, a set of agents with specialized expertise can be quickly assembled to help with the gathering of relevant information and knowledge, to cooperate with each other and with other parts of the production management system and humans to arrive at timely decisions in dealing with various enterprise scenarios. The proposed multiagent system, including its architecture and implementation, is presented and demonstrated through an example integration scenario involving real planning and execution software systems.

The interoperability problem: bringing together mobile agents and agent communication languages

Interoperability is a central issue far both the mobile agents community and the wider agents community. Unfortunately, the interoperability concerns are different between the two communities. As a result, inter-agent communication is an issue that has been addressed in a limited manner by the mobile agents community. Supplementing mobile agents with the ability to interact a a higher level with other mobile or static agents, or agentified information sources is a necessity in the vastly heterogeneous arena where mobile agents are called to compete. An agent communication language (ACL) has been interpreted by the agents community as a tool with the capacity to integrate disparate sources of information. Such a perspective does not seem to be shared by the mobile agents community. We investigate the origins of the differences of perspective on agent to agent communication and examine the reasons for the relative lack of interest on ACLs by mobile agents researchers and explore the integration of ACLs into mobile agents frameworks.

An agent-based infrastructure for enterprise integration

Some of the features that make Jackal (Java-based Applications Communication using KQML Agent communication Language) extremely valuable to agent development are its conversation management facilities, its flexible, blackboard-style interface and its ease of integration. Jackal has been developed in support of an investigation of the use of agents in enterprise-wide integration of planning and execution for manufacturing. This paper describes Jackal at a surface level and at the design level, and demonstrates its use in a multi-agent system that supports intelligent integration of enterprise planning and execution.

Coordinating agents using agent communication languages conversations

Internet agents are expected to accomplish their tasks despite heterogeneity; agents of different designs and of varying skills and domain knowledge need to interact successfully through knowledge and information exchange and effective coordination. We identify two distinct and separate problems that Internet agents are faced in an open and dynamic environment: knowledge sharing and coordination.