Web 服务合成多Agent系统综述

Web 服务合成多Agent系统综述
 
摘要：Web服务是1种新兴的Web应用种子，它们是自包含、自描述、模块化的应用，能通过Web 发布、定位和调用。Web服务技术的兴起，为建立1个高度分散化、高度自治分布的、适合Web环境的应用提供了计算模型。但是单个的Web服务功能有限，多个Web服务的有机组合能提供复杂功能以满足不同的Web用户需求。与单个Web服务相比，Web服务合成无疑会使得服务功能更强大，并且无需开发新的服务。目前Web服务合成的方式多为静态的和手工的，如果人工地从成千上万个Web服务中发现并选择最合适的Web服务进行组合，工作量将相当巨大。正是在这种压力的驱使下，众多的研究人员及研究团体投入到了自动化、智能化、精确化地搜索、匹配、合成Web服务的研究工作中。从Web服务合成研究的整体情况来看，Web服务合成领域的研究主要集中在下列几个方面：动态服务合成、服务合成质量和语义服务合成等。Web服务合成中最引人注目的几个技术分别是：基于QoS的合成、基于工作流的合成和基于本体的合成等。本文结合实例对它们分别进行了比较详细的论述。
Agent和多Agent系统是210世纪910年代出现的1种分布式计算模型。Agent是1类在特定环境下能感知环境，并能自治地运行以代表其设计者或使用者实现1系列目标的计算实体或程序。它的最基本特性包括反应性、自治性、面向目标和针对环境性等。多Agent系统是松散耦合的问题求解器网络，多个问题求解器协同工作来解决问题，其能力优于单个的问题求解器。这些问题求解器就是1个个的Agent。 多Agent系统的核心是“局部运作，全局共享”，可实现任务分担、自治、并发、合作、决策、通信和信息共享等。 （转载自科教范文网http://fw.nseac.com）
比较两者可以发现，Agent和Web服务实质上都是自描述、模块化、可重用的软件组件，它们在本质上的相似性说明这两种技术的结合是可行的。而且，它们都是未来主要的分布式计算模型。因此，Web服务合成与多Agent系统的结合是1种趋势，同时，多Agent系统为灵活地、自主地、智能地、自适应地Web服务合成提供了1种新的方法。
本文尝试结合多Agent系统理论，分析和探讨将其引入到Web服务合成的必要性和可行性，并综述了当前多Agent系统理论在Web服务合成应用中的若干实例。
关键词： Web服务合成；Agent；多Agent系统；工作流；本体
 
Survey on Web Services Composition with MAS
 
Abstract： Web services are a new breed of Web applications. They are self-contained, self-describing, modular applications that can be published, located, and invoked across the Web. The emerging of Web services provides a compute model for distribute applications, which are higher decentralization, independent distribution and suitable for Web environment.  However, single Web service has limited function, while multi Web services which are well structured, can provide complicated services to meet variable Web users’ needs. Compared with single Web service, Web service composition will doubtlessly strengthen its service function, furthermore, there is no need to explore new services. Currently most of Web service composite methods are static and manual. The workload will turn to be quite tremendous under such pressure, numerous researchers and research groups devote themselves to the automation, intelligibility and precision of the searching, matching and composition of Web services. Viewing from the whole situation, the research of the Web service composition domain includes the following aspects: the dynamic service composition, the QoS of the Web services composition, the efficiency of service composition and the like. The most profound technologies in the Web services composition include composition based on ontology, composition based on workflow, composition based on QoS component etc. In this paper we discuss these in detail combine with some examples. （转载自http://zw.ＮＳＥaＣ.com科教作文网）
Agent and Agent System emerged in the 1990s are a distribute compute model. Agent is a class of compute entity or program that can react with environment and run autonomously to represent designer or user to achieve a series of goal. Its basic characteristics include reactivity, autonomy, task-oriented, enacting to environment etc. Multi-Agent System is relaxed and loosely coupled problems solvers, which cooperating with tasks to solve problems, surpasses single problem solver. These problem solvers are single Agent. And the core of the Multi-Agent System can be described as following:“ All Agents can communicate each other and share the results which are figured out by every individual one”, which can realize task-sharing, autonomy and message-sharing. So we can see that Agent and Web services are self-describing, modularization, reusable in soft module in essential, as a result, the similarity shows that the combination of these two techniques is feasible. And then,Web Services Composition Multi-Agent System offers a method which can automatically, intelligently,  flexibly and self-suitably composite Web services.
In this paper, we attempt to combine the features of Multi-Agent System, introduce it into Web services composition. And the same time we have analyzed and discussed the necessity and feasible of such technology. Finally, we give a survey on a few examples of Web services composition making use of Multi-Agent System. 
Keywords:  Web Service;  Composition;  Workflow;  Ontology;  Agent;  Multi-Agent System (MAS)