A literature-based approach to annotation and browsing of Web resources

The emerging Semantic Web technologies critically depend on the availability of shared knowledge representations called ontologies, which are intended to encode consensual knowledge about specific domains. Currently, the proposed processes for building and maintaining those ontologies entail the joint effort of groups of representative domain experts, which can be expensive in terms of co-ordination and in terms of time to reach consensus. In this paper, literature-based ontologies, which can be initially developed by a single expert and maintained continuously, are proposed as preliminary alternatives to group-generated domain ontologies, or as early versions of them. These ontologies encode domain knowledge in the form of terms and relations along with the (formal or informal) bibliographical resources that define or deal with them, which makes them specially useful for domains in which a common terminology or jargon is not soundly established. A generalpurpose metamodelling framework for literature-based ontologies - which has been used in two concrete domains - is described, along with a proposed methodology and a specific resource annotation approach. In addition, the implementation of a RDF-based Web resource browser - that uses the ontologies to guide the user in the exploration of a corpus of digital resources- is presented as a proof of concept.