gistfile1.txt

Proposal: URIs can be used in several ways to relate a description to descriptions elsewhere. In some ideal cases, two descriptions may be clearly making assertions about the same entity. In this case, using identical URIs or mapping them with owl:sameAs can be useful. In many other cases, the relationship between described items (and each publisher's knowledge of the relationship) may be significantly less clear. It could be modeled in terms of topics (skos:Concept), types (rdfs/owl Class), and relationships could be treated with relations such as skos:exactMatch, skos:closeMatch, type instantiation (rdf:type), topical occurance (dc:subject). Granularity confusion due to levels of detail, changes over time etc. will be commonplace.The most appropriate terminology will vary by application area (e.g. polygon modeling for pathology tissue segmentation imagery) and other pragmatic factors (e.g. cost, knowledge of database contents etc.). In all cases, URIs provide an excellent notation for expressing any identifiers, regardless of precision and vocabulary choice. URI syntax is widely supported in many structured data syntaxes, and (via IRIs) has good support for I18N/L18N.