1.7 Hypermedia and Information Retrieval

Apparently the idea of combining IR and hypertext is as old as the hypertext idea itself. Making available all knowledge known to mankind in the Memex[2] as envisioned by Vannevar Bush [Bus45] demands extremely powerful information location capabilities. Since then, many efforts have been made to combine the two techniques, resulting in powerful systems combining the best of both worlds [Kah88]. One obvious combination of hypertext and IR ideas is the action link, where the link consists of an IR query. If users follow the link, they implicitly execute the query and get a graphical representation of the result set.

For the hypermedia system Intermedia [Yan88] (see also "Intermedia Web View" in chapter 7), a query sheet builder has been implemented that provides a graphic idea map, gives a view of the entire document collection, shows the location of the most important documents (with respect to the query) as well as giving access to the retrieved documents. Once a complicated query sheet has been assembled, it can be saved and later applied to other document collections, while still maintaining links to the original document collection. The result set of a query sheet can be saved as links to the actual documents, which can then be linked into the actual work environment of the user.

Chapter 4 introduces search systems for the web and Hyper-G such as WAIS and SWISH that are based on the IR concepts described in this chapter. Later in the book we present our own system for the exploration of large information spaces called CYBERMAP. CYBERMAP uses the IR techniques described in this chapter, in particular extensions of the vector space model and document clustering to give an overview and provide the navigation means to browse and explore large hypermedia documents.