Discovery of Semantic Structures and Access Control in Distributed Hypermedia

在这项研究中，我们开发了新的高精度信息检索方法和一种动态访问方法，以更有效地在万维网(WWW)等分布式超媒体上传播信息。我们特别关注以下研究课题，并开发了评估这些算法所需的算法和原型：(1)从Web数据中发现语义结构及其在分类和信息组织中的应用为了从Web页面中发现主题结构，我们开发了一种算法，分别根据术语的出现密度分布和术语的共现关系来发现主题术语和相关术语。基于这些算法，我们可以成功地发现网页之间的泛化和聚集关系。我们还开发了一个原型系统，用户可以根据他们的主题结构来检索具有相关反馈的Web数据。(2)基于重构和同步检索数据的自动生成我们开发了一种个性化和生成同步内容的新方法和原型。通过多次信息传递实验验证了该方法的有效性。(3)基于用户浏览历史和偏好的Web数据动态访问控制我们开发了一种个性化网页的方法和原型，该方法使用基于规则的XML语言来表示Web作者的意图。(4)基于新鲜度和本地度的非传统Web信息过滤

In this research, we developed new information retrieval methods with high precision and a dynamic access method for more effective information dissemination for distributed hypermedia such as World Wide Web (WWW). We especially focused on the following research topics, and developed necessary algorithms and prototypes for evaluating those algorithms :(1) Discovery of Semantic Structures from Web Data and Their use for Classification and Information OrganizationIn order to discover a topic structure from a Web page, we developed an algorithm to discover subject terms and related terms by the term appearance density distribution and the term co-occurrence relationships, respectively. Based on these algorithms, we could successfully discover the generalization and aggregation relationships among Web pages. We also developed a prototype system in which users can retrieve Web data with relevance feedback by their topic structures.(2) Automatic Data Generation Based on Restructuring and Synchronizing Retrieved DataWe developed a new way and a prototype for personalizing and generating synchronized contents. The effectiveness of the proposed method was evaluated through several information delivery Experiments.(3) Dynamic Access Control of Web Data by User Browsing Histories and PreferencesWe developed a method and a prototype for personalizing Web pages, which uses a rule-based XML language for representing Web author's intensions.(4) Non-traditional Web Information Filtering Based on the Freshness and the LocalnessWe also developed new information filtering methods based on discovery of the freshness and the Localness of Web pages