Concept naming vs concept categorisation: a faceted approach to semantic annotation

The purpose of this paper is to show that concept naming alone in document annotation is not sufficient to convey the thought content of the information resource. The paper presents an outline of semantic document annotation which combines two major processes: facet analysis and concept categorisation. This is also an effort to show how RDF schema can be designed and implemented so that the properties of the schema are able to express the basic structure of the subject matter of the resource.

This paper presents a methodology for representing the subject matter of a document in terms of RDF. For the purposes of faceted subject annotation, it has developed an extended RDF schema for simple knowledge organisation system (SKOS). The facets and relationships of the faceted subject indexing language postulate‐based permuted subject indexing system (POPSI) have been transformed into RDFS classes. The elementary categories of POPSI form the property classes in the POPSI/RDF Schema. These property classes have been used to formulate the subject description of a document.

The subject annotation of a document using this schema expresses all the components of the thought content of an information resource.

The examples given in this paper show the applicability of this schema in describing resources in web directories and annotating scholarly documents in digital libraries. In a broader perspective, this provides a methodology for formulating the subject metadata of web resources. This schema helps in formulating the subject string(s) for a resource outlining the skeleton structure of its thought content.

SKOS has been developed as an RDF schema representation of the traditional knowledge organisation systems. But the schema has limited room to accommodate subject indexing languages. The present schema extends the SKOS schema to accommodate the representation of faceted subject indexing languages. The faceted subject annotation system has been adopted for the very reason that it has precedence over the enumerated classification systems, controlled vocabulary lists, etc. The potential to describe the specific subject of the document with more accuracy and representation of context gives the faceted subject indexing languages strength to make the subject description explicit and machine processible.