定向查询引擎在Web化学数据库集成检索中的应用

Internet上的化学数据库是重要的专业资源，基于超链接分析的搜索引擎还不能索引这类资源。本论文以充分利用Internet上的化学数据库数据为目标，将“一个查询发动多个同级检索引擎，并以结构化的方式组织信息”的方案应用于以化合物标识信息为检索入口的Web化学数据库，建立了一个基于多站点集成检索的Web数据库定向查询引擎。该引擎是一个包括用户交互层、中间检索层、数据提供层的三层Web模型。各层在系统内部分别对应于响应用户检索请求的客户端代理模块、集成远程Web信息的服务器端代理模块，以及提供缓存和检索的关系数据库模块。模型采用JSP＋Java组件的开发方式，在HTTP协议标准发送方法的基础上，采用XML技术对检索返回文档进行结构化数据的提取和表示，利用XML—DBMS实现XML数据的存储和检索，建立了一套针对深层Web数据交换的解决方案。依此方案所建立的ChemDB Portal Search实现了四个分布式Web化学数据库的有效加入、同时检索和统一显示。该系统是针对深层Web信息的挖掘和集成检索的一次尝试，它可为其它领域建立类似的系统提供借鉴。

Directed query engine applications in the integrated retrieval of chemical Web databases

The data in Internet Chemical databases are a class of valuable resources, which couldn't be indexed by search engines based on hyperlink analysis. The major purpose of this paper is to take good advantage of these resources. This is an approach that one query launches several search engines at host sites of distributed chemical databases with compound identifications as entry points in a cascading fashion, and searching results organized in a structural way to form a ChemDB Portal Search, a Chemical Directed Query Engine. A three-tier model is designed for the approach. , including the user interface as a Client Agent responding users'queries, the searching middle-tier ware as a Server Agent integrating data from the target sites, and the Web sites and local database as the data managers providing retrieval of the data. Combining with HTTP to send queries, the model is implemented with JSP + JavaBean fashion using XML technology to wrap structural data from the returned pages and XML-DBMS to store and retrieve XML documents in local databases. Simultaneous searching of five distributed chemical databases by one query is now possible to the ChemDB Portal Search, which can display the hits from different sources in a unified form. In conclusion, the thesis is an attempt to mine and integrate data from Deep Web. It may provide a practicable approach for building similar systems in other fields.