Vast amounts of text are now available in machine-readable form and can be processed electronically. The vector space model of text processing has been widely used and has consistently produced superior retrieval results for the last thirty years. Traditionally , information retrieval research has concentrated on improving on-demand retrieval of useful textual information. Often a user has no particular information need; instead, he is just interested in browsing the available information. Most hypertext systems depend on manually created browsing structures to support text browsing. Manual text linking does not scale up to the amount of electronic information accessible in today's networked computing environment. The vector space model can be used to automatically impose a exible structure on arbitrary texts. Tools can then be provided for eeective browsing of this structured text.
Download Full PDF Version (Non-Commercial Use)