Page 116 - JOURNAL OF LIBRARY SCIENCE IN CHINA 2018 Vol. 42
P. 116
WU Wenna & BAO Xiulin / The architecture and data model of the National Thesauri Warehouse 115
Concepts are independent and traceable resources in the web environment with URIs as
unique identifications. The property “skos:inScheme” describes which concept schemes the
concepts belong to. Three combinations of properties are used to describe literal forms, i.e.
labels of concepts. Preferred labels (preferred terms) of concepts use the property combination:
skosxl:prefLable, skosxl:literaForm. Non-preferred labels (non-preferred terms) of concepts use the
property combination: skosxl:altLable, skosxl:literaForm. Hidden concept labels use the property
combination: skosxl:hiddenLabel, skosxl:literalForm. Misspellings of terms are treated as hidden
labels. Misspellings are rarely used as non-preferred terms in existing thesauri. Nevertheless,
considering practical application, it’s helpful to increase recall ratio to take in some misspellings
which appear frequently.
There are two types of links that connect concepts. 1) Links between concepts in same
concept scheme: concepts can be interconnected through the property “skos:semanticRelation”
and its sub-properties “skos:broader”, “skos:narrower” and “skos:related”. 2) Links between
concepts in different schemes: concepts can be linked cross schemes through the property
“skos:mappingRelation” and its five sub-properties. Cross-scheme includes multiple cases:
between different schemes in a thesaurus (for example, between main table and classification);
between different thesauri (for example, between a NTW’s member thesaurus and the CT);
between a member thesaurus and the top ontology, etc.
The property “skos:notation” is used to describe category, and references category codes.
Mapping relationships, for example, skos:broadMatch, are used to define the categories a common
concept belongs to.
5.2 Metadata scheme for term description
Terms are labels of concepts (skosxl:label). In NTW, terms are described as resources. SKOS-XL
data model provides two extended properties “skosxl:literalForm” and “skosxl:labelRelation”, to
describe terms. The property “skosxl:literalForm” is used to describe literal forms of terms. Every
term has one and the only one literal form. The property “skosxl:labelRelation” is used to describe
relations between terms. In traditional thesauri, relations between terms are semantic relations.
However, in the SKOS-XL data model, terms are not connected through semantic relations (see
4.1). The SKOS-XL data model offers only one property “skosxl:labelRelation” to describe
relations between terms. But, if needed, this property can be extended to meet special description
demands. For example, the relationships between correct spellings and misspellings (misSpellOf,
misSpell); the relationships between full names and abbreviations (abbreviationOf, abbreviation).
The SKOS-XL data model lacks properties to describe phonetic notation. Thus, the property
“skos:note” should be extended to get its sub-property “phoneticNotation” which would be defined
and used to describe phonetic notations of terms. The metadata scheme for term description is