An intelligent quasi-thesaurus for accessing information systems

Abstract

The information resources available through the Internet are immense. The total volume of the files accessible via the Internet is counted in thousands of gigabytes. The main problem however on the Internet is the flood of information and the difficulty of simply finding items. Therefore, the issue of identifying the relevant resources and accessing them is of crucial importance, in particular from the casual users' perspective. As a rule users' specific knowledge about the network itself, and the distribution and contents of the informàtion sources is limited. This paper addresses this issue by proposing a method helping the users to establish "good" queries and submitting them to the information resources (residing on the network) which are likely to contain the relevant documents. The core idea ofthe proposed approach is to create a self-Iearning mechanism supporting the queries' establishment and forwarding them to the right places within the network. The mechanism, called ADDRESSER, has been conceived as a simple quasi-thesaurus which is composed ofterms, URL addresses related to the terms and relations linking the terms. A prototype for testing and evaluating the idea was implemented. The experiments have proved the idea is viable, sound and workable.