摘要
随着网络通信技术的发展以及数据库应用规模的扩大，传统的单机数据库系
统已正逐渐向物理多节点环境下的多数据库系统发展。越来越多的数据信息通过
本地的局部缓存机制，被物理分布在网络的各个节点上。因此，如何将分散在不
同网络节点上多个数据源连接成为一个有机的整体，以实现更为广泛的资源共
享，成为当前研究的热点问题。
本文提出了一种基于MAS技术进行多个异构数据源集成的逻辑模型，该模
型通过多个agent间的交互，达到了异构多数据源集成的目的。最后，本文在此
逻辑模型的基础上实现了一个原型系统，验证了该模型的可行性。本文的创新点
主要体现在以下几个方面：
1．使用MAS技术实现异构多数据源的集成，使得集成系统可获得较高的透
明性；
2．集成系统不但实现了异构数据源间的互访，而且可在系统中形成统一的数
据操纵、管理界面（SDI）；
3．通过相应的功能扩充，系统可支持半结构化/非结构化数据源的集成，扩大
了数据共享程度；
4．集成系统可有效支持冗余服务，提高了系统的可靠性和可用性。
As development of networks and enlarging scale application of databases, the
traditional databases on single machine are gradually switching to multi-database
system on multi-node. More and more information has been distributed to different
nodes in networks through the mechanism of local cache. So, it becomes the research
focus that how to integrate multi data sources distributed in different network nodes
into one system and improve the degree of resource shared.
A logical model of integration of multi heterogeneous data sources based on
MAS is introduced in this paper. Multi heterogeneous data sources can be integrated
by the interaction of multi agents in the model. At the end, a prototype is implemented
based on the logical model to prove that the model is feasible. The created works of
the paper are listed as follows:
a. The technique of MAS is used to implement the integration of multi
heterogeneous data sources. It can make the system obtain the higher
transparency.
b. The system can not only provide the method to access all the data sources
transparently, but also provide a uniform interface (SDI) for user to
manipulate the data and manage the system.
c. The system can support the integration of semi-structured/non-structured data
sources. The range of information shared is enlarged.
d. The system can support redundant services. The reliability and the availability
of the system are improved.
引文
[1]Shoham Y. Agent-Oriented programming. Artificial Intelligence. 1993, 60(1):51～92.
[2]Shoham Y. & Tennenholtz M. Computational Social Systems: offline design.Artificial Intelligence, 1995, 73 (1-2): 231～252.
[3]Boutilier C., Shoham Y. & Wellman M. P. Economic Principles of Multi-Agent Systems (editorial), Journal of Artificial Intelligence 1997, 94(1-2): 1～6.
[4]Wooldridge M. J. & Jennings N. R. Intelligent agent: theory and practice.Knowledge Engineering Review, 1995, 10 (2).
[5]李毅等．多agent系统的一种交互策略模型．软件学报，1999，10(7)．
[6]Dunin-Keplicz B, Verbrugge R. Collective commitments. In: Proceedings of the 2nd International Conference on Multi-agent System. Tokyo: 1996.56～63.
[7]朱晓芸等．一种基于智能Agent的协同工作模型．计算机应用，1999，10(6)．
[8]周永林，潘云鹤．面向Agent的分析与建模．计算机研究与发展，1999，36(4)：410～416．
[9]徐泽同．分布式数据库，工程设计CAD及自动化，1998，4．
[10]陈建中等，智能Agent建模的一种模板结构．计算机研究与发展，1999，36(10)．
[11]毛新军等，Agent在多Agent系统中计算的意愿理论．软件学报，1999，10(1)．
[12]刘大有等．Agent研究现状与发展趋势．软件学报，2000，11(3)：315～321．
[13]焦文品，史忠植．构造MAS的动态体系结构的模型．计算机学报，2000，23(7)：732-737．
[14]董红斌，石纯一．一个多Agent系统模型．小型微型计算机系统，2000，21(7)：740～743．
[15]李保建等．一种基于类构架的agent程序设计方法，小型微型计算机系统，1999，20(2)．
[16]Carlos A. Iglesias, MIX: A General Purpose Multi-Agent Architecture,Proceedings of IJCAI'95.
[17]Jean-Francois Arcand, Cognition Based Multi-Agent Architecture, Proceedings of IJCAI'95.
[18]Evandro de Barros Costa, A Multi_Agent Interactive Learning Environment Model, Proceedings of ICCE97
[19]何炎祥等著．高级操作系统．科学出版社，1999．
[20]朱海滨等著．分布式系统原理与设计．国防科技大学出版社，1997．
[21]李建中，孙文隽著．并行关系数据库管理系统引论．科学出版社，1998．
[22]钱方．面向冗余服务的分布对象模型．国防科技大学博士论文，2000．
[23]周巍松等著．Linux系统分析与高级编程技术．机械工业出版社，1999．
[24]陈莉君著．Linux操作系统内核分析．人民邮电出版社，2000．
[25]http://www. gnu.org/manual/manual.html. GNU Manuals Online-GNU Project-Free Software Foundation (FSF).
[26]http://www. mysql.com/documentation/index.html. MySQL Documentation.
[27]何欣枫等．智能代理技术在网络集成中的应用．小型微型计算机系统，2001，22(5)：556～558．