Page 109 - JOURNAL OF LIBRARY SCIENCE IN CHINA 2018 Vol. 42
P. 109

108 Journal of Library Science in China, Vol. 8, 2016


            as parent tables while main tables, auxiliary tables and classifications as child tables. Between
            the two types of tables, there are relationships of “including” (dct:hasPart) and “being included”
            (dct:isPartOf). Concepts in main tables reference categories in classifications, so relationships
            between main tables and classifications are “referencing” (dct:references) and “being referenced”
            (dct:referencedBy). As for thesauri with multiple editions, different editions can be regarded as
            different concept schemes. Relationships between different editions are “replacing” (dct:replaces)
            and “being replaced” (dct:isReplacedBy).

                                  5IFTBVSVT
                                QSFWJPVT FEJUJPO
                                               EDU JT3FQMBDFE#Z
                                       EDU SFQMBDFT
                                                  5IFTBVSVT
                                                              EDU IBT1BSU
                                         EDU IBT1BSU
                                EDU JT1BSU0G               EDU JT1BSU0G
                                                                  "VYJMJBSZ
                               .BJO 5BCMF    EDU JT1BSU0G          5BCMF "
                                    EDU SFGFSFODFE#Z  EDU JT1BSU0G EDU IBT1BSU
                                EDU SFGFSFODFT   EDU IBT1BSU  "VYJMJBSZ
                                                              5BCMF
                                         $MBTTJGJDBUJPO


                                        *OEFY "  *OEFY

                                       Figure 2. Thesaurus macrostructure.


            3.2  Thesaurus metadata


            Thesaurus metadata is a description scheme for tables of thesauri. Thesaurus metadata is very
            similar to that of a document. Therefore, it’s feasible to design a metadata scheme of NTW based
            on the metadata scheme of Dublin Core (Dublin Core Metadata Initiative, 2008), as shown in
            Table 1.
              In NTW system, member thesauri and their constituent parts can be described as independent
            concept schemes. Their “ResourceType” (rdf:type) is “conceptScheme” (skos:conceptScheme).
            According to specific demands the class “conceptScheme” may be further divided into “thesaurus”
            “mainTable” “classification” “auxiliaryTable” and so on. All concept schemes on different levels
            use URIs as unique identifications. Although main tables, classifications and auxiliary tables can
            be described respectively, nonetheless, in most cases, they are not granted with independent table
            names. Thus, when describing, they inherit and extend the names of related thesauri, for example,
            “Chinese Thesaurus_Classification”. The disciplines, which a thesaurus belongs to or covers,
            can be described with property “dc:subject”, and its values reference categories in the NTW top
            classification.
   104   105   106   107   108   109   110   111   112   113   114