Requirements for components of an intelligent information retrieval model for the WWW

We are currently witnessing an eruption of (potentially) useful information that is made available over the Internet. Providing easy and efficient access to this information is crucial bearing in mind the high levels of distribution and semantic heterogeneity (compounded also by syntactic mismatches) exhibited by the Web. The paper describes a framework for information retrieval based on knowledge sharing and reuse. The authors adopt a common three-layer WWW architecture, consisting of a user agent, an information location service provider (mediator), and a resource access mechanism (wrapper). Their contribution is to functionally decompose these components in order to give more precise meaning to their roles, and to enable design decisions about the division of labour (and associated meta-knowledge) between them. They set out requirements for an adequate information resource description so that one can handle the variety of information resources connected to the Internet. They show that it is desirable for the meta-knowledge to be communicated between the WWW layers and propose that it should be part of the mediator functionality.