Towards supporting user-defined hypertext contexts in Web searching

A hypertext context is a generic high-level hypermedia structure that groups together a set of nodes and links into a logical whole. It can be used as boundaries of searches in large hypertext systems and helps to improve the results of searches concerning specific topics or subject domains. On the current Web, there exist no hypertext contexts that are explicitly represented and can be recognized automatically. However, a search system may provide facilities to help users to form their mental context views and transfer the views to concrete instantiations of hypertext contexts, and then enables users to search in these contexts. This paper describes such a prototype system, with which users call specific hypertext contexts by selecting seed pages and including their descendants and/or ancestors within a certain distance, adding single pages, and excluding pages flexibly when they search and browse on the Web. This work reveals a significant potential to improve Web searching by making use of high-level hypermedia structures.