中文摘要：

阐述了一种应用于网络检索的多Agent框架，该框架有3种Agent组成：管理Agent用于提取信息的语义和管理；接口Agent为用户和系统提供接口；搜索Agent用于在WWW上搜索信息。本工作的重点是使用XML技术在wwW上检索信息。在这个模型中，Agents间使用XMI。作为Agent通信语言，相互沟通执行检索任务。它们的知识，语义以及搜索结果都用XML格式来表示。用户不仅能更精确地从模型搜索结果的语义编码中获取信息，而且可以直接运用搜索结果，无需使用专用标签或专门的脚本来获得网页，从中提取内容。

英文摘要：

A multi-Agent frame for network retrieval is introduced. This frame consists of three types of Agent： managing Agent for extracting the semantics and management of information; the interface Agent for providing interface between the system and users; the search Agent for searching information on WWW. This work focuses on the use of XML technology for information retrieval on WWW. In this model, XML was used as Agent communication language with each other to perform retrieval tasks, their knowledge, semantics and search results were expressed by XML format. As a result, users not only can access information more precisely from semantically encoded search results that are returned from the model, but also can uti lize the content of the results without using proprietary tags or customized scripts to scrape web pages and to extract the content.