INFORMATION-RETRIEVAL FROM HYPERTEXT - AN APPROACH USING PLAUSIBLE INFERENCE

This paper approaches the problem of information retrieval from hypertext. In this context, the retrieval process is regarded as a process of inference that can be carried out either by the user exploring the hypertext network, browsing, or by having the system exploit the hypertext network as a knowledge base, searching. In the following, a comprehensive model is proposed taking into account both of the perspectives, and combining effectively browsing and querying into a unified framework. Hypertext nodes are regarded as facts, links as rules, and the connected hypertext structure as an inference network that can be used to prove the query inferentially. Next, design and implementation issues are discussed concerning a prototype system developed to evaluate the proposed approach in terms of retrieval effectiveness and search efficiency.