Constructing a Virtual Digital Library for Browsing Conceptually Structured Document Collection

参考传统图书馆的活动，我们开发了一种虚拟数字图书馆（一种信息系统），在这种系统中，任何用户都可以通过浏览系统结构化的文档集合来满足他/她的信息需求。数字图书馆的文档是通过用户配置文件过滤的Internet中的文本数据，这是关于用户兴趣的描述列表，并且是预先安排的，因此概念上相关的文档被赋予类似（但不同）的整数。由于与查询在概念上相关的文档可以通过这种编号形成集群，因此编号过程对应于通常的库活动中看到的索引任务。当用户提出他/她的查询（将给出最小的数字）时，上述文档编号将被重新排列，以便在概念上与查询相关的文档被给予尽可能小的数字。虚拟数字图书馆是通过将查询和文档簇放置在一个距离空间上，其坐标反映上述文档编号来构建的。用户可以通过浏览靠近查询的集群和/或包含与查询相关文档的集群来满足他/她的信息需求。为了方便用户的浏览，设计并实现了一个面向VRML/JAVA的WWW导航器。这里提出的浏览功能比通常检索引擎中广泛使用的排序功能更有效。

We have developed by referring to the activities in a traditional library a virtual digital library (a kind of information systems), in which any user can satisfy' his/her information needs by browsing a systematically structured document collection. The documents for a digital library are the textual data in the Internet filtered through a user-profile, a list of descriptions about user's interests, and are pre-arranged such that the conceptually related ones are given similar (but different) integer numbers. Since the conceptually related documents to the query can form clusters by this numbering, the numbering process corresponds to the indexing task seen in the usual library activities.When a user poses his/her query (which will be given the smallest number) the above document numbering is re-arranged such that the documents conceptually related to the query are given small numbers as much as possible. A virtual digital library is constructed by placing the query and the document clusters on a distance space, where their coordinates reflect the above document numbering.The user can satisfy his/her information needs by browsing the clusters placed close to the query and/or to the clusters containing the relevant documents to the query. To assist the browsing process, a VRML/JAVA oriented navigator on WWW is also designed and realized. The browsing capability proposed here is shown to work more effectively than the ranking capability widely utilized in the usual retrieval engines.