【作者】 何莉；

【导师】 顾君忠；

【作者基本信息】 华东师范大学 ， 计算机应用技术， 2007， 硕士

【摘要】 互联网的出现极大地丰富了人们的信息来源。然而由于缺乏统一的组织和管理，人们在浩瀚的信息海洋中却难以找到所需信息。当前各种信息服务技术，如搜索引擎、基于Web Service的服务集成等，要么精度不高、返回结果过多，要么仅局限于提供Web Service接口的信息资源、覆盖范围有限，难以满足人们日益增长的信息需求。鉴于Web上绝大部分信息资源以半结构化形式存在的现状，本文在研究Web信息抽取技术的基础上，提出了一种“基于Web信息抽取的个性化信息服务”框架WINIS(Web INformation extraction based personalized Information Service)，主要着眼于传统搜索引擎难于发现、隐藏在Web站点后台数据库中的丰富信息(也称为暗藏网)，尝试通过Web信息抽取技术获取该类资源，从而为用户提供个性化的信息服务。基于WINIS，用户无需关心信息的来源和获取的方式，只需描述自己的需求即可简单、高效、快速地获得目标信息。可以将WINIS简述如下：在框架的个性化信息服务层，系统通过定义任务模式来描述和解析用户请求，并提出基于用户目标的个性化结果整合策略。在框架的Web信息抽取层，采用一种基于本体的Deep Web信息抽取方法来获取暗藏网中的信息资源，在保证抽取质量的同时大大减轻了用户负担，有效解决了现有方法中用户负担大、缺少待抽取页面获取过程、抽取结果缺乏语义信息等问题。以WINIS框架为指导，本文构建了E-Planning原型系统并进行了实验分析。分析结果表明，基于WINIS框架的E-Planning系统在信息抽取质量、结果方案生成以及系统扩展性等方面都达到令人满意的结果，从而验证了本文提出的WINIS框架的有效性。更多还原

【Abstract】 The Internet enriches sources of information for people. However, due to the heterogeneity and lack of structure of Web information resources, people often feel very difficult to find information they need from Internet. Current systems for information service, like search engines and Web Service integration systems, often have limitations in accuracy and coverage, which can not satisfy pepople’s growing needs for information.At present, the lager amount of information on the Web is stored in HTML documents.In this dissertation, we firstly analyze the Web information extraction technology, and then present WINIS, a framework for web information extraction based personalized information service. Aimed at the rich information hidden in the database of web site(also called the Deep Web), WINIS retrieves it by Web information extraction and provides personalized information service to end users, for whom it is not necessary to know where and how to get the information they need.WINIS is composed of Personalized Service Layer and Web Information Extraction Layer. In the Personalized Service Layer, predefined task shema is used to decompose user’s request and a user goal based method for answer composition is proposed. In the Web Information Extraction Layer, an ontology based approach for Deep Web information extraction is presented to retrieve information from Deep Web. This approach extracts data from Deep Web sources effectively and automatically and it solves the key problems of current approaches, such as too much interaction with users, lack of semantic support and process for getting the pages.By using WINIS framework as guidance, we built a prototype system of E-Planning. The experiments show that WINIS based E-Planning system performs well in data extraction, result plan generation and also in expansibility, which indicate the validity of WINIS framework.更多还原